114 NO114 DECEMBER • 2014

Aspects of recent developments in the Latin American and Caribbean labour markets Jürgen Weller 7

The earnings share of total income in Latin America, 1990-2010 Martín Abeles, Verónica Amarante and Daniel Vega 29

Latin America: Total factor productivity and its components Jair Andrade Araujo, Débora Gaspar Feitosa and Almir Bittencourt da Silva 51

Financial constraints on economic development: Theory and policy for developing countries Jennifer Hermann 67

CEPAL REVIEW CEPAL The impact of China’s incursion into the North American Free Trade Agreement (nafta) on intra-industry trade Jorge Alberto López A., Óscar Rodil M. and Saúl Valdez G. 83

Work, family and public policy changes in Latin America: Equity, maternalism and co-responsibility Merike Blofield and Juliana Martínez F. 101

A first approach to the impact of the real exchange rate on industrial sectors in Colombia Lya Paola Sierra and Karina Manrique L. 119

Global integration, disarticulation and competitiveness in Mexico’s electromechanical sector: A structural analysis Raúl Vázquez López 135 REVIEW Technological capacity-building in unstable settings: Manufacturing firms in Argentina and Brazil

CEPAL Anabel Marín, Lilia Stubrin and María Amelia Gibbons 153 ECONOMIC Index of political instability in Brazil, 1889-2009 COMMISSION FOR Jaime Jordan Costantini and Mauricio Vaz Lobo Bittencourt 171 LATIN AMERICA AND THE CARIBBEAN

ISSN 0251-2920 LATIN AMERICA AND COMMISSION FOR THE CARIBBEAN ECONOMIC ECONOMIC

REVIEW

NO114 DECEMBER • 2014

Alicia Bárcena Executive Secretary Antonio Prado Deputy Executive Secretary REVIEW

ECONOMIC COMMISSION FOR LATIN AMERICA AND THE CARIBBEAN

Osvaldo Sunkel Chairman of the Editorial Board André Hofman Director Miguel Torres Technical Editor

ISSN 0251-2920 The cepal Review was founded in 1976, along with the corresponding Spanish version, Revista cepal, and it is published three times a year by the United Nations Economic Commission for Latin America and the Caribbean (eclac), which has its headquarters in Santiago, Chile. The Review, however, has full editorial independence and follows the usual academic procedures and criteria, including the review of articles by independent external referees. The purpose of the Review is to contribute to the discussion of socioeconomic development issues in the region by offering analytical and policy approaches and articles by economists and other social scientists working both within and outside the United Nations. The Review is distributed to universities, research institutes and other international organizations, as well as to individual subscribers.

The opinions expressed in the signed articles are those of the authors and do not necessarily reflect the views of eclac. The designations employed and the way in which data are presented do not imply the expression of any opinion whatsoever on the part of the secretariat concerning the legal status of any country, territory, city or area or its authorities, or concerning the delimitation of its frontiers or boundaries.

To subscribe, please visit the Web page: http://ebiz.turpin-distribution.com/products/197587-cepal-review.aspx

The complete text of the Review can also be downloaded free of charge from the eclac web site (www.eclac.org).

This publication, entitled the cepal Review, is covered in the Social Sciences Citation Index (ssci), published by Thomson Reuters, and in the Journal of Economic Literature (jel), published by the American Economic Association

United Nations publication ISSN: 0251-2920 ISBN: 978-92-1-121881-7 e-ISBN 978-92-1-057211-8 LC/G. 2629 Copyright © United Nations, December 2014. All rights reserved. Printed in Santiago, Chile

Requests for authorization to reproduce this work in whole or in part should be sent to the Secretary of the Publications Board. Member States and their governmental institutions may reproduce this work without prior authorization, but are requested to mention the source and to inform the United Nations of such reproduction. In all cases, the United Nations remains the owner of the copyright and should be identified as such in reproductions with the expression “© United Nations 2014”. a r t i c l e s

Aspects of recent developments in the Latin American and Caribbean labour markets 7 Jürgen Weller

The earnings share of total income in Latin America, 1990-2010 29 Martín Abeles, Verónica Amarante and Daniel Vega

Latin America: Total factor productivity and its components 51 Jair Andrade Araujo, Débora Gaspar Feitosa and Almir Bittencourt da Silva

Financial constraints on economic development: Theory and policy for developing countries 67 Jennifer Hermann

The impact of China’s incursion into the North American Free Trade Agreement (nafta) on intra-industry trade 83 Jorge Alberto López A., Óscar Rodil M. and Saúl Valdez G.

Work, family and public policy changes in Latin America: Equity, maternalism and co-responsibility 101 Merike Blofield and Juliana Martínez F.

A first approach to the impact of the real exchange rate on industrial sectors in Colombia 119 Lya Paola Sierra and Karina Manrique L.

Global integration, disarticulation and competitiveness in Mexico’s electromechanical sector: A structural analysis 135 Raúl Vázquez López

Technological capacity-building in unstable settings: Manufacturing firms in Argentina and Brazil 153 Anabel Marín, Lilia Stubrin and María Amelia Gibbons

Index of political instability in Brazil, 1889-2009 171 Jaime Jordan Costantini and Mauricio Vaz Lobo Bittencourt cepal Review referees 2013 and January-August 2014 191

Guidelines for contributors to the cepal Review 193 Explanatory notes The following symbols are used in tables in the Review: … Three dots indicate that data are not available or are not separately reported. (–) A dash indicates that the amount is nil or negligible. A blank space in a table means that the item in question is not applicable. (-) A minus sign indicates a deficit or decrease, unless otherwise specified. (.) A point is used to indicate decimals. (/) A slash indicates a crop year or fiscal year; e.g., 2006/2007. (-) Use of a hyphen between years (e.g., 2006-2007) indicates reference to the complete period considered, including the beginning and end years. The word “tons” means metric tons and the word “dollars” means United States dollars, unless otherwise stated. References to annual rates of growth or variation signify compound annual rates. Individual figures and percentages in tables do not necessarily add up to the corresponding totals because of rounding. Aspects of recent developments in the Latin American and Caribbean labour markets

Jürgen Weller

ABSTRACT This article presents three stylized facts that characterized the evolution of labour markets in Latin America and the Caribbean in the period 2003-2012 and represented breaks from previous trends. It is argued that these changes have to do with the economic and production context and the political and institutional framework. We show how the magnitude and patterns of economic growth impact on the nature of job creation, especially on shifts within and between economic sectors and the various segments of different productivity levels. We emphasize how changes in labour institutions have contributed to the evolution of labour indicators and, lastly, look at recent advances and persistent weaknesses in labour performance, as well as a number of risks to the continuity of recent favourable labour trends.

KEYWORDS Employment, labour market, economic growth, employment creation, unemployment, labour productivity, income, employment statistics, Latin America, Caribbean region

JEL CLASSIFICATION J21, J46, J88

AUTHOR Jürgen Weller is Chief of the Employment Studies Unit of the Economic Development Division of the Economic Commission for Latin America and the Caribbean (eclac). [email protected] 8 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

In the first decade of the 2000s, labour performance in the economic-industrial and political-institutional diverged markedly from the previous two decades in setting. Section II sets forth some stylized facts with Latin America and the Caribbean. Especially from regard to labour in the recent period, and contrasts 2003 onwards, there was a break from several previous them with previous performance. Section III presents trends: the unemployment rate came down, many a scheme developed to analyse the factors determining employment quality indicators improved amid increasing these changes, justifying the emphasis on the economic- job formalization, and wage gaps between skilled and industrial and political-institutional context. Section unskilled workers narrowed. These factors helped to IV examines the characteristics of these factors in the reduce poverty rates and inequality between households decade 2003-2012, furthering understanding of recent in this period. developments in the labour market. Here, we show This favourable picture may be attributed in part that the relatively high economic growth in recent times to a number of factors that weighed in to one extent was the main factor driving job creation and bringing or another in the countries of the region, especially down the regional unemployment rate. A modest rise economic and production factors and political-institutional in labour productivity, thanks to several processes, developments. facilitated gains in employment quality, as did strong The analysis in this article centres on the 10-year labour demand, which led to many of the new jobs being period beginning in 2003, looking at how a more benign created in medium- and high-productivity segments and economic and production setting affected the labour shifted the structure of employment favourably. Patterns markets. By contrast with the two preceding decades, of labour demand also helped to reduce income gaps in this period the region’s economy enjoyed relatively between people at different levels of employment. Lastly, sustained, high growth rates, interrupted only by the a shift in labour institutionality in many countries also economic and financial crisis of 2008-2009.1 At the same helped to close gaps and drove gains in employment time, many countries moved away from the growth and quality. Section V concludes by discussing recent progress development paradigm that prevailed in the region in the and persistent weaknesses in labour performance, and 1980s and 1990s (reliance on the superior efficacy and draws attention to a number of factors that pose risks to efficiency of the functioning of markets), which implied, the continuity of recent benign labour trends.2 in many cases, adopting new labour policies. As this article will discuss, both factors had major consequences for labour market performance in this period. The sections following this introduction review 2 This article does not attempt to analyse all aspects of the recent recent changes in the region’s labour markets and labour market. The review of stylized facts concentrates instead on aspects that show a break in trend from previous performance, and interpret them against the backdrop of the changes skips over indicators that remained the same. This refers, for example, to the characteristics of labour supply, in which demographic trends held steady, labour participation continued to rise gradually and level of formal education continued to increase. Obviously, these stylized facts 1 See eclac (several years) and, specifically for the subperiod 2003- are not applicable to all the countries of the region, since they each have 2008, Kacef and López-Monti (2010). specific patterns regarding the determinants of labour performance.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 9

II Three stylized facts

The first stylized fact regarding the development of The fall in unemployment starting in 2003 resulted labour markets over the recent period is the reduction from a rapidly rising employment rate, as shown in in the open unemployment rate in urban areas. Between figure 1. 2002 and 2012 this rate came down from 11.2% to 6.4% While the participation rate reflected the long-term at the regional level, to reach a 20-year low (eclac/ilo, gradual increase in women’s participation in the labour 2013).3 This was a break from the general upward trend in market, the employment rate rose sharply as of 2003, unemployment in the 1990s, when the rate fell slightly in interrupted only in 2009 by the global economic and times of high economic growth and rose significantly financial crisis. in times of zero or slack growth, producing an upward- Starting in 2003, the faster rise in employment, sloping jagged pattern in the regional unemployment rate.4 together with higher wages for those in work, pushed up labour income, which was the main driver of poverty reduction in the region in this period (eclac, 2012, 3 The rate thus returned to the levels of the 1970s and 1980s. However, p. 50; Azevedo and others, 2013). a precise comparison is not possible because of methodological changes made in many countries in the intervening period. The second stylized fact is the improvement 4 In a study on 19 Latin American countries for the period 1990-2007 in employment quality. As noted in Weller and —although with differences depending on data availability— Ball, De Roethlisberger (2011), over the recent period almost all Roux and Hofstetter (2011) found evidence of a hysteresis effect that quality indicators show improvements, except for a rise transformed short-term rises in unemployment (caused by contractions in aggregate demand) into long-term impacts. in temporary contracts, which was attributable to more

FIGURE 1 Latin America and the Caribbean: urban participation, employment and unemployment, 1990-2012 (Percentages)

62

60

Participation rate 58

56 Unemployment

54 Employment rate Proportion of the working-age population

52

50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Employment rate Participation rate

Source: prepared by the author, on the basis of official data from the countries. Note: in the figure unemployment is shown as the gap between the participation and employment rates, so the values do not correspond to the unemployment rate, which is calculated as a proportion of the workforce.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 10 CEPAL REVIEW 114 • DECEMBER 2014 unstable labour markets. Quality indicators tend to be the wage gap was the main driver of reduction in income better for wage earners than for non-wage earners, but inequality at the household level in the past decade. For an in the recent period employment quality has in general average of 15 countries, around two thirds of the reduction improved for both categories.5 came from smaller differences in income per employed As noted in the work cited above, these improvements person (eclac, 2012, p. 56).7 The second important are closely linked to processes of labour formalization factor was non-labour income, with the introduction which have made great strides in some countries. As and expansion of programmes targeting the poorest shown in figure 2, in the decade 2003-2012 formal households. By contrast, demographic changes in the employment expanded more than in the preceding period, great majority of the countries (a smaller gap between and notably more than employment overall, in many of the dependency rates of the first and fifth quintiles) the region’s countries. contributed very little to the lessening of inequality. On The third stylized fact is the reduction in wage gaps, average, the contribution of higher employment rates in a break from the previous trends of wage structure —which had a positive impact on poverty reduction— polarization.6 In most of the countries, the reduction in had even a slightly negative effect on equality because the proportion of employed adults rose at least as much or more in the higher income quintiles as in the lowest. 5 See information on coverage differentiated by type of social protection system and changes in this regard in ilo (several years). 6 On wage gap narrowing, see, for example, Gasparini and others (2011); López-Calva and Lustig (2010); and Keifman and Maurizio (2012). 7 In this connection, see also Azevedo, Inchauste and Sanfelice (2013).

FIGURE 2 Latin America (selected countries): annual growth in (registered) formal employment and total employment, 1990-2002 and 2003-2012 (Percentages)

8

7

6

5

4

3

2

1

0

-1 Chile Brazil Mexico Panama Uruguay Argentina Nicaragua Guatemala Costa Rica El Salvador

Formal employment 1990-2002 Formal employment 2002-2012 Total employment 2002-2012

Source: prepared by the author, on the basis of official data from the countries. Note: the data on formal employment refer to those paying into or affiliated to a system of contributory social protection, except in Brazil (register of formal employment). The data on total employment for Argentina and Uruguay refer to urban employment. The data on total employment for Guatemala and Nicaragua refer to 2002-2011 and 2003-2010, respectively. In the first period, the data on formal employment refer to 1991-2002 for Panama; 1994-2002 for Chile, Mexico and Nicaragua; 1995-2002 for Argentina, and 1998-2002 for El Salvador.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 11

III Conceptual aspects

The main determinants of employment generation and an economy’s average labour productivity. This affects job characteristics are the economic and production the capacity to improve job quality, because it is a context and labour institutions.8 Figure 3 summarizes determinant of the profit margin allocated to workers. In the main links in this regard. this respect, the main link is between labour productivity The production context is expressed, first, in and wage level. The comparison between countries shows economic growth based largely on greater use of that this link tends to be a close one, although labour production factors, including labour. Growth impacts productivity gains do not translate automatically into not only the magnitude of employment generation, but wage gains, as demonstrated by the recent worsening of also —through its characteristics (sectoral composition, functional income distribution.9 But non-wage aspects technical intensity, and so forth)— labour demand, of employment quality tend to benefit from increased differentiated by skills, knowledge, etc., and therefore productivity too, inasmuch as the various aspects of quality influences the relative labour conditions of different tend to generate costs, and the capacity of economies groups of workers. to cover those costs depends on the resources available. Second, and obviously related to the first aspect, Accordingly, high productivity tends to feed directly into the economic and production context is also expressed in good quality employment, while low productivity usually

8 Here, we are leaving aside the close long-term correlation between employment level and demographic evolution (magnitude and growth of the working-age population), modified by the upward trend in labour participation caused by the increasing incorporation of women 9 eclac/ilo (2012) shows both the high correlation between labour into the labour market. In this connection, see Weller and Kaldewei productivity and average wages, and the recent fall in wages as a (2013, pp. 31 and 32). proportion of gdp.

FIGURE 3 Determinants of employment generation and quality

Economic factors Quality factors Institutional factors

Characteristics of the labour force (human capital)

Production context Standards on safety, (economic growth, health and productivity and hygiene at work heterogeneity) Quantity and quality of employment Other production factors (capital, Standards on technology and labour relations, others); organization training of production

Source: prepared by the author, on the basis of Jürgen Weller and Claudia Roethlisberger, “La calidad del empleo en América Latina”, Macroeconomía del Desarrollo series, No. 110 (LC/L.3320-P), Santiago, Chile, Economic Commission for Latin America and the Caribbean (eclac), 2011. United Nations publication, Sales No. S.11.II.G.39.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 12 CEPAL REVIEW 114 • DECEMBER 2014 goes hand in hand with processes of exclusion from of both formality and informality, it must be emphasized productive employment.10 that there are theoretical and empirical reasons to treat Rising labour productivity at the aggregate level markets as segmented (Fields, 2004; Ocampo, Rada and reflects two different, but related, dynamics. On the Taylor, 2009; Infante, 2011): one hand, economic development is closely related to —— The region’s production structure is highly structural change processes that involve reassigning heterogeneous, with large productivity gaps between resources from low-productivity sectors —which and within sectors. typically create low quality jobs— to intermediate- and —— Because of weak labour demand in medium- and high-productivity sectors (productivity gain through high-productivity segments and the jobs that are intersectoral changes). On the other hand, productivity created mainly in response to this demand, other with an activity can be boosted by more intensive use jobs are created mainly in response to the pressure of physical or human capital or both, by technological of labour supply, reflecting the need for income in changes, by more efficient use of resources, or by the the respective households.11 closure of unproductive firms and the formation of new In any case, the different approaches in the literature firms that are more productive (productivity gain through on labour market segmentation have made progress in intrasectoral changes). The relative weight of the two recognizing the heterogeneous nature of the informal processes typically varies with the march of economic sector itself.12 While certain approaches differentiate development, with the contribution of structural change between “excluded” and “voluntary” informal workers, gradually diminishing (Rodrik, 2013). from the standpoint of production, a subsegment with a However, and in third place, in highly heterogeneous certain capacity of accumulation is distinguished from production structures, the importance of the production one whose income is basically stuck at subsistence level. context in generating and informal employment cannot Accordingly, there would appear to be a countercyclical be captured by the rate of economic growth and the subsegment —which grows in phases of the economic evolution of average productivity alone. cycle in which the labour demand from the medium- and The segmentation hypothesis is often refuted by the high-productivity segments is low— and another that argument that a significant proportion of employment behaves procyclically, expanding in response to income informal is voluntary and arises from cost-benefit opportunities in growing economies. calculations on the part of firms and workers. Specifically, In addition, the simultaneous existence of different it is possible to tap into the social benefits that are also labour market segments —whose behaviour is determined increasingly accessible for informal employees, without chiefly by labour supply and demand, respectively— is having to assume formalization-related costs. Studies evident in the way labour adjusts to the economic cycle. on flows of workers within this market also indicate a Specifically, the fact that a large segment of the labour degree of mobility between the different segments, which market obeys labour supply dynamics contributes much appears to contradict the theory of segmented markets to the highly procyclical behaviour of labour productivity (Perry and others, 2007; Bosch, Cobacho and Pagés, (eclac/ilo, 2012). 2012). Without denying the role of cost-benefit rationale In sum, the level and composition of informality depend on two main factors: the production structure —represented by per capita gdp and the productivity 10 Weller and Roethlisberger (2011, pp. 54-58) show some correlations between per capita gdp (as an indirect variable for labour productivity) gaps between the different segments— and the legal and and several indicators of employment quality. The most obvious positive links occurred between payment into social security systems (pensions and health), contracts and, to a lesser extent, bonuses and vacation. Other indicators (trade union membership, overtime pay, 11 training) showed high dispersion, which points to the importance In this article “segment” means the set of those parts of different of other factors beyond per capita gdp and, related to that, average branches of activity or sectors that share similar productivity levels, productivity, specifically the importance of labour institutions, which drawing a simplified distinction between low-productivity segments will be examined later. Notably, the link between labour productivity and medium- and high-productivity segments. The shift in the relative and employment quality is two-directional, as has been observed importance of the two segments over the business cycle is recognized historically, for example with the shortening of the working day. by Perry and others (2007). Efficiency wage theory examines the impact of better wages on higher 12 See, for example, Tokman (1987); Fields (2004) and Perry and productivity. In this regard, see Akerlof and Yellen (1986). others (2007).

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 13 institutional factors that determine the costs and benefits in this regard (for example, through formalization or of both formality and informality.13 minimum wage policies). Labour institutions have a The existence of segmented markets with large dual function (eclac, 2010, p. 173): (i) contribute to the productivity gaps means that changes in the composition efficient functioning of the labour market (by bringing of occupational structure —especially in the relative the workforce increasingly into productive jobs), boost weight of the lower- or higher-productivity segments— productivity (including the effective distribution of the influences its overall averages as well as labour income. fruits thereof) and design mechanisms to adjust to the Labour market segmentation has distributive implications fluctuations of the economic cycle; and (ii) promote as well. Although the theories disagree as to whether the protection of workers as the structurally weakest low-productivity segments should be modelled with stakeholders, especially vulnerable groups with specific marginal output close to zero or positive, declining problems in participating in productive labour. output (Fields, 2004), an increase in the number of To this end, labour legislation and collective individuals employed in these segments will tend to bargaining affect the process of establishing wages and lower average income and widen the income gap with other benefits, contracts, social protection, and training, respect to the higher-productivity segments. By contrast, among other aspects. In turn, standards on hygiene and a reduction in the number of individuals in these poor safety at work and on the organization of work (pace, performing segments will increase average income and breaks and so on) affect labour conditions. narrow the gap with respect to the average or higher- On the other hand, the capacity of firms to comply, productivity segments. One implication of this is that the coverage and efficiency of labour surveillance and the income gap between different productivity levels labour justice, and the control and pressure that workers will be wider in countries with a larger set of low- can exert determine the degree to which the quality productivity segments than in countries with a smaller aspects covered in the legislation or negotiations come set of such segments: in the first group of countries, to fruition (Bensusán, 2008). low-productivity segments tend to show smaller average Institutions can only meet their objectives in a income, owing to the larger proportion of jobs generated sustainable manner if they are positioned on two fronts by supply pressure, thereby increasing the gap between (Berg and Kucera, 2008, p. 27): first, they reflect the this income level and the income of medium- and social norms created historically in a specific country, high-productivity segments.14 so they vary from one country to another. It is thus no Lastly, labour institutions —based on labour surprise that the theoretical literature is increasingly legislation or on collective bargaining— shape the considering the possibility of moving from models that characteristics of employment, particularly matters of assume a single optimum configuration of regulations, quality and the gaps between different groups of workers to others that allow for the existence of two or more institutional configurations capable of generating similar outcomes (Eichhorst, Feil and Braun, 2008). Second, 13 There are also secondary aspects, such as the lack of information they respond to the production context, so that in any on these costs and benefits for informal firms and workers. given country they can vary over time in response to 14 Weller (2012, p. 35) demonstrates this for Latin America. social and political changes.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 14 CEPAL REVIEW 114 • DECEMBER 2014

IV Changes in the determining factors and in their links with the stylized facts

This section explores how changes in the economic and gross fixed capital formation).17 By sector, growth was production context and in labour institutions may have concentrated in the various branches of the tertiary sector, impacted the stylized facts described in section II during and this pattern deepened between 2009 and 2012.18 the period beginning in 2003. As shown in figure 4, the rise in the employment rate noted earlier (see figure 1) is attributable in large 1. The economic and production structure measure to these higher rates of economic growth. With the exception of 2009, in all the years between (a) Economic growth and overall employment 2003 and 2012 the urban employment rate rose by 0.4 The growth and characteristics of employment were percentage points or more. the main factor explaining the rise in employment levels, The correlation between economic growth and and thus the declining unemployment rate. In the period employment rate variation indicates, as well, an increase beginning in 2003, the Latin American and Caribbean in the labour intensity of growth. The data in figure 4 economies evolved very differently from the preceding show that in the 1990s, economic growth of 3% at the decades. New global growth patterns and buoyancy pushed regional level was accompanied by a drop in the regional up prices for the region’s main export commodities, urban employment rate of 0.1 percentage point, while in which improved terms of trade to the benefit of countries the period beginning in 2003 this same output growth exporting oil, gas, minerals and agricultural products.15 rate was accompanied by a rise of over 0.3 point in This was a key factor in boosting the region’s economic the employment rate (eclac, 2014b). The low labour growth above the performance of previous decades. intensity of growth in the 1990s has been associated, Economic authorities used the opportunity offered by among other causes, with the structural reforms of the this context to reduce vulnerabilities, lower public debt time (Weller, 2000; idb, 2003). This effect could have (especially external debt), and build up international eased later, and in the following period there were few reserves. The combination of brighter growth prospects large-scale liberalizing reforms (eclac, 2014b). Be this as and reduced vulnerabilities helped to improve external it may, the low labour intensity of growth at the regional borrowing conditions, especially amid highly liquid level in the 1990s chiefly reflected the performance of global markets (eclac, several years). a few countries (Argentina and Brazil), and in most The region’s economic growth rose from 2.7% per of the countries of the region “jobless growth” was year between 1990 and 2002 to 3.8% per year from 2003 not the rule; most of them were rather characterized to 2012, despite the harsh impact of the global economic at this time by “employment without growth” (Pagés, and financial crisis in 2008-2009.16 Growth patterns have Pierre and Scarpetta, 2009), reflecting the supply side varied throughout the past decade, since in 2003-2008 pressure of labour in the context of low and volatile exports increased rapidly (from 16.9% of gdp on average economic growth. in 1990-2002 to 22.7% on average in 2003-2008, at In sum, by comparison with the previous period, constant prices). By contrast, in 2009-2012 household in the decade 2003-2012 the employment rate was consumption and gross fixed capital formation were the significantly boosted by higher economic growth and main drivers of aggregate demand (with average rises stronger intensity of the labour factor in the process. from 63.0% in 2003-2008 to 65.0% in 2009-2012 for Since the increase in the participation rate did not household consumption and from 19.1% to 21.2% for diverge from its long-term trend, the more robust job

15 Differences in the composition of the export and import baskets 17 Author’s calculation on the basis of data from the Economic were obviously reflected in very varied terms of trade evolutions in Commission for Latin America and the Caribbean (eclac). the different countries of the region. 18 Together, these branches of activity increased their share in gdp 16 The region grew by 4.5% per year between 2003 and 2008, 4.4% from 54.5% on average for 1990-2002 to 55.3% for 2003-2008, and per year between 2010 and 2012, and contracted by 1.6% in 2009. 56.8% for 2009-2012.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 15

FIGURE 4 Latin America and the Caribbean: economic growth and variation in the urban employment rate, 1991-2012

1

0.5

0 -3 -2 -1 0 1 2 3 4 5 6 7

-0.5

Variation in the employment rate -1

-1.5 Economic growth (percentages)

1991-2002 2003-2012 Linear trend (1991-2002) Linear trend (2003-2012)

Source: prepared by the author, on the basis of data from the Economic Commission for Latin America and the Caribbean (eclac). creation translated directly into a notable fall in the intersectoral change, for the period 1990-2005, also in unemployment rate. the average for nine countries.20 Table 1 shows the results of an exercise for the (b) Changes in labour productivity periods 1990-2002 and 2003-2011/2012, for which Labour productivity evolution is a key determinant the methodology used by McMillan and Rodrik (2011) of improvement in both wage and non-wage aspects of was applied to 23 countries of Latin America and the employment quality. In 2003-2012, the more benign Caribbean, as follows: economic environment led to a rise in average labour productivity, which had fallen in the 1980s and stood n n DHYy=+DDy H still in the 1990s (Weller and Kaldewei, 2013).19 ti//,,tk- it it,,it i ==11i As noted earlier, this increase could have come from changes within economic sectors, or from where Yt and yi,t represent the level of productivity the reallocation of resources —specifically the for the economy overall and for sector i, respectively, workforce— between sectors (structural change). while Θi,t is the share of sector i in employment. Δ For the 1980s, eclac (2007) found a sharp fall in represents the change in productivity or in the proportion average labour productivity, caused by intrasectoral of employment, as the case may be. The first term on the effects with slightly positive intersectoral effects, right is the sum of the variations in productivity of the while for 1991-2003 it found —in the simple average different sectors, weighted by their share in employment for nine countries— a slight rise in average labour at the start of the period under analysis. This term thus productivity, with small positive contributions from represents changes in productivity within sectors. The both components. By contrast, McMillan and Rodrik second term represents the contribution of structural (2011) found a strong positive contribution from change to total productivity variation, calculated as the intrasectoral changes, partially offset by negative sum of sectoral changes in share of total employment, weighted by the corresponding productivities.21

19 However, the 1.6% rise in average labour productivity between 20 2002 and 2012 was modest by comparison with some other world For other recent studies that break down the evolution of labour regions. Owing chiefly to the rapid increase in productivity in East productivity for different periods, see Ocampo, Rada and Taylor Asia, by 2012 the gap that the region had gained with respect to the (2009) and Ros (2011). world average, which expanded by 2.2% in this period, was almost 21 Thus, if a sector loses share in the structure of employment to closed (author’s calculation on the basis of data from Key Indicators another sector with higher average productivity, aggregate productivity of the Labour Market (kilm), eighth edition (ilo, 2013)). increases, and vice versa.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 16 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 1 Latin America and the Caribbean: variation in average labour productivity and contributions of intra- and intersectoral changes, 1990-2002 and 2003-2011/2012

1990-2002 2002-2011/2012

Period Yearly Yearly Intersectorial Intrasectorial Intersectorial Intrasectorial productivity productivity contribution contribution contribution contribution variation variation

Argentina urban areas a 1990-2002 and 2002-2012 1.0 0.8 0.2 3.9 0.2 3.7 Bahamas 1989-2003 and 2003-2011 0.0 0.4 -0.4 0.2 0.3 -0.1 Barbados 1990-2002 and 2002-2012 -1.1 -0.7 -0.4 1.8 0.3 1.5 Bolivia (Plurinational 1996-2002 and 2002-2009 1.5 0.1 1.4 0.2 1.4 -1.2 State of) Brazil 1990-2002 and 2002-2011 -0.4 0.0 -0.4 1.9 0.7 1.2 Chile 1990-2002 and 2002-2012 3.4 -0.2 3.6 1.4 0.5 0.9 Colombia 1991-2000 and 2002-2012 1.0 -0.3 1.3 1.7 0.8 0.9 Costa Rica 1990-2002 and 2002-2012 0.7 0.3 0.4 2.3 0.1 2.2 Dominican Republic 1991-2002 and 2002-2012 2.9 0.1 2.8 1.8 -0.1 1.9 Ecuador 1990-2001 and 2002-2012 -0.7 -0.1 -0.6 3.5 0.8 2.6 El Salvador 1992-2002 and 2002-2012 1.4 1.3 0.1 0.3 0.1 0.3 Guatemala 1989-2002 and 2002-2011 -0.3 -0.7 0.5 1.8 2.0 -0.2 Honduras 1990-2002 and 2002-2012 -0.4 0.5 -0.9 1.7 0.2 1.6 Jamaica 1992-2002 and 2002-2012 0.0 0.2 -0.2 -0.2 0.3 -0.5 Mexico 1991-2002 and2002-2012 0.3 0.9 -0.5 0.8 1.1 -0.3 Nicaragua 1990-2003 and 2003-2010 -1.0 0.5 -1.5 -0.7 -0.1 -0.6 Panama 1991-2002 and 2002-2012 0.2 0.9 -0.8 4.8 0.3 4.5 Paraguay 1997-2002 and 2002-2011 -2.6 -0.6 -2.0 0.5 0.2 0.3 Peru 1994-2002 and 2002-2011 -1.7 -0.4 -1.2 4.0 1.7 2.3 Saint Lucia 1994-2002 and 2002-2007 -0.6 -0.3 -0.3 -0.2 1.1 -1.2 Trinidad and Tobago 1990-2002 and 2002-2012 1.0 0.4 0.6 3.2 0.4 2.8 Uruguay urban areas a 1990-2002 and 2006-2011 1.8 0.9 0.9 4.2 0.7 3.4 Venezuela (Bolivarian 1990-2002 and 2002-2012 -2.9 -1.6 -1.3 1.5 1.7 -0.2 Republic of)

Latin America and the 0.2 0.1 0.1 1.8 0.6 1.1 Caribbean b Northern subregion of 0.5 0.5 0.0 1.6 0.4 1.2 Latin America b Southern subregion of 0.1 -0.1 0.2 2.3 0.9 1.4 Latin America b The Caribbean b -0.1 0.0 -0.1 1.0 0.5 0.5

Source: prepared by the author, on the basis of Jürgen Weller and Cornelia Kaldewei, “Empleo, crecimiento sostenible e igualdad”, serie Macroeconomía del Desarrollo, No. 145 (LC/L.3743), Santiago, Chile, Economic Commission for Latin America and the Caribbean (eclac), 2013; and official data from the countries and the Economic Commission for Latin America and the Caribbean eclac( ). a The data for Argentina (31 urban agglomerations) and Uruguay (for 1990-2002) (urban total) are indicative only, because information on non-agricultural output growth was combined with urban employment data, in the absence of nationwide employment figures. b Simple average. Note: rates of intra- and intersectoral contributions do not necessarily add up to the variation in productivity, because of rounding. The total refers to the sum of value added of the different branches of activity. The calculation was performed in dollars at constant 1995 prices for 1990-2002 and at constant 2005 prices for 2002-2011/2012. The northern subregion corresponds to the countries from Mexico to Panama, plus the Dominican Republic; the southern subregion corresponds to the Latin American countries to the south of Panama.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 17

The result of this calculation shows the contribution the patterns of employment generation in segments with of intrasectoral processes and structural change to variation different productivity levels, as will be discussed in the in aggregate labour productivity, in constant dollar at following section. 1995 prices for the first subperiod and at 2005 prices for the second. In order to compatibilize the results of TABLE 2 the countries individually, and in view of the fact that Latin America and the Caribbean (23 countries): coefficient of variation the periods under analysis are of different length, annual of labour productivity in the different growth rates of labour productivity were calculated branches of activity for all countries and the contributions of intra- and (Averages) intersectoral changes were translated into contributions to those growth rates. Weighted average Simple average Again, average labour productivity growth was 1990 0.78 0.96 faster in the second period, as compared with small 2002 1.05 1.04 rises in the first. In the regional average, between 1990 2002 1.63 1.15 and 2002 neither intrasectoral changes nor structural 2010-2011 1.36 1.08 change contributed to average labour productivity Source: Jürgen Weller and Cornelia Kaldewei, “Empleo, crecimiento growth, whereas both made a positive contribution in sostenible e igualdad”, Macroeconomía del Desarrollo series, the second period, with structural change representing No. 145 (LC/L.3743), Santiago, Chile, Economic Commission for about a third of aggregate labour productivity growth, Latin America and the Caribbean (eclac), 2013. and intrasectoral changes the other two thirds. There were Note: the first two lines (1990 and 2002) are based on productivity calculations in dollars at constant 1995 prices; the following two large differences between countries and subregions in (2002 and 2010-2011) are based on productivity calculations in both periods, however. In the first period, intersectoral dollars at constant 2005 prices. The column “Weighted average” changes made a stronger contribution in the northern refers to the coefficient of variation of productivity of the various branches in the weighted average for the countries. subregion, possibly because of the expansion of the maquila industry and mass emigration, especially to the United States. In the second period, the southern In sum, in 2003-2012, process of intra- and subregion showed better results in both components intersectorial change generated modest rises (with of productivity growth. In both periods, the Caribbean significant differences between countries) in labour showed weaker results. productivity. This supported gains in real wages and Slightly rising labour productivity helped bring improvements in non-wage indicators of employment. about increases in real wages. According to household survey data, on average in the Latin American countries, (c) Economic growth and employment generation in average urban wages rose from 4.1 to 4.7 poverty lines the segments of production per capita between 2002 and 2012, having risen from As proposed in section III, in heterogeneous labour 3.8 to just 4.2 poverty lines per capita between 1990 and markets it is not enough to review the impact of changes 2002 (eclac, 2013). However, real wages generally rose in aggregate indicators between and within branches of at lower rates than labour productivity, which points to activity; it is also essential to look at changes occurring the importance of the work of labour institutions, as between and within segments of varying productivity well (eclac/ilo, 2012). levels. This section illustrates the importance of this The evolution of labour productivity differentiated second aspect.22 by branch of activity, resulting from intra- and As shown in figure 5, in the 1990s urban employment intersectoral changes, also affected productivity gaps generation was concentrated mainly in low-productivity between these branches. As may be observed in table segments, which —in the simple average for the countries 2, in 1990-2002 productivity gaps broadened between with information available— increased their share of branches of activity in the region (both in the weighted average of productivities, and in the simple average of productivities at the level of each economic activity), 22 Since the information available is not sufficient to measure the size thereby deepening the region’s hallmark structural of these segments regularly, labour indicators are used as a proxy. For some time now, eclac has been measuring the size of low-productivity heterogeneity. By contrast, in the following period the segments in urban areas by the proportion of employed who are own- gaps narrowed by about the same amount as they had account workers (excluding professionals and technical workers), widened in the previous period, reflecting changes in wage earners and employers in microenterprises, domestic service workers, or unpaid family workers.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 18 CEPAL REVIEW 114 • DECEMBER 2014 urban employment from 45.7% to 50.1%. Starting segments driven by supply and demand, respectively: around 2002, the proportion of employment in medium- wage employment and own-account work.23 and high- productivity segments rose in almost all the Figure 6 shows, for the region overall, a high positive countries in the region, and fell in the low-productivity correlation (0.86) between economic growth and wage segments to around 44.1% on average. Owing to data availability limitations, the analysis of the relationship between economic growth and the 23 Although not all wage employment corresponds to the demand- driven segment, and not all own-account work to the supply-driven characteristics of employment generation looks at segment, as will be seen, the great majority of the first category shows two categories of employment as proxies for labour demand-driven patterns, and most of the second, supply-driven patterns.

FIGURE 5 Latin America: proportion of the urban employed working in low-productivity segments, 1990-2002 and 2002-2012 (Percentages)

70 70 65 65 60 60 55 55 50 50 45 45 200 2 2012 40 40 35 35 30 30 25 25 20 20 20 30 40 50 60 70 20 30 40 50 60 70 1990 2002 Source: Economic Commission for Latin America and the Caribbean (eclac), Social Panorama of Latin America, 2013 (LC/G.2580), Santiago, Chile, 2013. United Nations publication, Sales No. E.14.II.G.6. la: Latin America.

FIGURE 6 Latin America and the Caribbean: economic growth and changes in employment generation, 1995-2012 (Percentages)

6

5

4

3

2 Growth rate

1

0

-1 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

-2 Wage employment Own-account work  Source: prepared by the author, on the basis of official information from the countries and the Economic Commission for Latin America and the Caribbean (eclac). gdp: gross domestic product.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 19 employment in 1995-2012, since wage employment rose level was -0.23 (in other words, the countercyclical notably in years of high economic growth and very little factor prevailed). in slow-growth or crisis years. These same results are found at the country level. There is, conversely, a much less clear-cut As shown in table 3, in the median for 14 countries relationship between economic growth and own-account wage employment has a correlation coefficient of 0.58 work. In several years, own-account work behaved with economic growth, whereas the coefficient for own- countercylically, reflecting supply-side dynamics (for account work is -0.27. Albeit with large differences example in 1995, 1996, 1999 and 2009 in the context between countries, wage employment behaves clearly of weak wage employment generation, and between procyclically. By contrast, the presence of both procyclical 2005 and 2007, in a relatively lengthy period of buoyant and countercyclical dynamics in own-account work leads wage job creation). This behaviour reflects the nature to quite low correlation coefficients; the negative sign of low-productivity segments, driven from the supply in almost all the countries appears to indicate that the side by household need. countercyclical pattern prevails, however. However, own-account work also behaves Higher economic growth thus especially favoured procyclically in some years, both amid relatively wage employment generation, which contributed to labour strong economic growth (1997, 2000, 2004, 2008 and formalization. Since the pattern of wage employment is 2010) —when not only was labour demand high, but closely correlated with economic growth, in the aggregate individuals needing income found favourable opportunities growth is a stronger determinant of total employment for independent work— and amid low growth (2001 generation in countries with higher relative rates of wage and 2002). employment24 (Weller, 2012). These varied dynamics underscore the internal The output elasticity of wage employment was heterogeneity of low-productivity segments, as approximately 0.5 in the median for the group of countries mentioned in section III. Not only those excluded from in the period 1995-2012. Although with much variation wage employment enter own-account work; there is between countries, the predominant pattern points to also a dynamic subsegment of opportunity-seekers. consistent efforts to increase productivity, which (in the The overall result of the different rationales was that in 1995-2012 the correlation between gdp growth and the generation of own-account work at the regional 24 Proportion of wage employment in total employment.

TABLE 3 Latin America (selected countries): employment-economic growth correlation coefficients andgdp elasticity of wage employment, 1995-2012

Employment-economic growth correlation coefficient gdp elasticity of Total employment Wage employment Own-account work wage employment Argentina (17) 0.71 0.77 -0.07 0.56 Brazil (18) 0.63 0.62 0.21 0.75 Chile (17) 0.54 0.65 -0.27 0.38 Colombia (18) 0.15 0.53 -0.34 0.26 Costa Rica (18) 0.37 0.45 -0.27 0.59 Dominican Republic (18) 0.61 0.20 0.50 0.54 Ecuador (17) -0.19 -0.13 -0.01 1.32 El Salvador (16) 0.04 0.44 -0.30 0.45 Honduras (14) -0.31 0.20 -0.03 0.91 Mexico (17) 0.79 0.87 -0.58 0.50 Panama (18) 0.34 0.70 -0.42 0.37 Peru (13) 0.08 0.33 -0.08 0.26 Uruguay (11) 0.67 0.77 -0.30 0.06 Venezuela (Bolivarian Republic of) (18) 0.47 0.78 -0.35 1.05 Latin America (median) 0.42 0.58 -0.27 0.52

Source: prepared by the author, on the basis of official information from the countries and the Economic Commission for Latin America and the Caribbean (eclac). Note: the figure in brackets is the number of years with information available in each country. gdp: gross domestic product.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 20 CEPAL REVIEW 114 • DECEMBER 2014 aggregate) is partially offset by the expansion —albeit The concentration of a large share of new jobs in slowly more recently— in low-productivity jobs, reflecting the medium- and high-productivity segments, reflecting persistent structural heterogeneity. strong labour demand, was related to the expansion of These changes in employment generation patterns formal employment in the tertiary sector, which in less helped to increase average labour income. Specifically, favourable circumstances typically provides an informal the concentration of new job creation in medium- or and low-productivity refuge amid labour supply pressure high-productivity segments led to a rise in average labour (eclac, 2014b). income, as shown in figure 7. As figure 8 shows, in many countries commerce While in the 1990s the expansion of employment and services, and also construction, showed rates of mainly in low-productivity segments had a slightly formal employment growth well in excess of the rate in negative impact on average labour income, in the manufacturing and agriculture. This sectoral breakdown following period part of the rise in labour income of formal employment reflects above all the pattern of was attributable to the reallocation of a portion of the economic growth increasingly focused on domestic- workforce from low-productivity segments to medium- demand-related activities (those producing mainly or high-productivity segments. non-tradable goods and services), which concentrated Since labour arrangements are much more formalized labour demand in those activities, thus facilitating the in the employment categories that make up the medium- expansion of formal employment there. and high-productivity segments, the shift in composition The largest gaps between the expansion of total reflects improvements in formality levels and, thus, in employment and formal employment occurred in the employment quality.25 agricultural sector, mining, commerce, and community, social and personal services. This appears to be due to three factors: first, some of these sectors have low entry 25 See Weller and Roethlisberger (2011) and ilo (several years). In barriers, so that in low growth periods supply-driven addition, formalization policies targeting both sectors may have an impact (in this respect, see section IV.2.b). employment creation is concentrated there, and this pattern

FIGURE 7 Latin America (median of 16 countries): breakdown of increases in labour income, by intra- and intersectoral contributions of segments of varying productivity levels, 1990-2002 and 2002-2011 (Per capita poverty line equivalents)

0.8

0.6

0.4

0.2

0.0 Intrasectoral Intersectoral Intrasectoral Intersectoral

2002-0991 1102-2002 -0.2

-0.4

-0.6

Latin America Southern subregion Northern subregion

Source: prepared by the author, on the basis of Economic Commission for Latin America and the Caribbean (eclac), Social Panorama of Latin America, 2012 (LC/G.2557-P), Santiago, Chile, 2013. United Nations publication, Sales No. E.13.II.G.6.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 21

FIGURE 8 Latin America (median of 11 countries): formal and total employment and value added by branch of activity, annual growth rates, 2002-2012 (Percentages)

8.0

7.0

6.0

5.0

4.0

3.0

2.0

1.0

0.0 Total Commerce Construction Financial and Manufacturing business services Transport, storage Community, social hunting and shing and communications Agriculture, forestry, and personal services Mining and quarrying Electricity, gas and water

Formal employment Total employment Value added Source: prepared by the author, on the basis of official data from the countries. Note: the information covers the following countries: Argentina, Brazil, Chile, Costa Rica, Ecuador, El Salvador, Guatemala, Mexico, Nicaragua, Panama and Peru. The data on formal employment refer to those paying into or affiliated to a contributory social protection system, except in Brazil (register of formal employment), Ecuador and Peru (surveys of firms with 10 or more workers in selected branches of activity). The data of annual growth in total employment and value added refer to the period 2002-2012, except in the cases of Brazil, Guatemala and Peru (2002-2011), Ecuador (2003-2012) and Nicaragua (2003-2010). eases when economic growth is higher. Second, formal The labour policies applied during this period were activities expanded strongly in some sectors, especially another important factor (see section IV.2). commerce, partly to the detriment of informal segments. There are also signs that the skills bias in labour And, third, efforts to formalize labour markets may demand may have reversed during the first decade of have had a differentiated effect by sector, with greater the 2000s. progress in the abovementioned sectors. Among the causes of narrower income gaps, In sum, the characteristics of labour demand have López-Calva and Lustig (2010) identify, for the four impacted the recomposition of employment, allowing case studies they present on Argentina, Brazil, Mexico advances in wage and non-wage aspects of work and and Peru, decreasing demand bias for skilled workers, improvements within branches of activity. as the impact of intensive technological change on qualifications (triggered by the economic reforms of (d) Growth, characteristics of labour demand and the 1980s and 1990s) runs its course. reduction of wage inequality Gasparini and others (2011) argue that demand for Did economic growth and its characteristics play a less skilled labour may have increased in the context of role in the third of the stylized facts, the narrowing of the commodity expansion and intrasectoral processes, such wage gap? As noted in section IV, the rise in the quantity as technology spread and the mismatch between the skills of employment did not contribute directly to reducing of workers with higher levels of formal education and inequality at the household level, but employment traits the jobs available. The World Bank (2012) suggests that may have narrowed the gap in labour income, the most the concentration of growth in non-tradable sectors may important mechanism in this regard. One key finding have reduced the demand for higher-skilled workers, was that the rise in educational level appears to have which is lower in these sectors than in manufacturing. helped to improve distribution by reducing gaps between Table 4 shows the composition of additional household quintiles with different levels of per capita employment for 15 countries, by level of education income (Cruces, García Domenech and Gasparini, 2012). and differentiating wage employment (reflecting

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 22 CEPAL REVIEW 114 • DECEMBER 2014 demand for labour on the part of firms) from non-wage In sum, labour demand continued to favour workers employment (representing different dynamics, mainly with an intermediate or high level of education, although supply-side pressure).26 As found for the 1990s (Weller, there were also new opportunities for those with little 2000), here, the majority of the new jobs are created are formal education. wage employment for workers with medium and high To analyse the sectoral characteristics of labour levels of education. As well, in keeping with progress demand, a breakdown was performed using the in education systems, only a very small percentage methodology of Berman, Bound and Griliches (1994) of new employment (around 5% in the median for 15 to estimate the contributions of intra- and intersectoral countries) corresponds to individuals with 9 or fewer changes both to higher demand for individuals with years of schooling. intermediate and high levels of formal education, and to Comparison of these results with those of the lower demand for individuals with less formal education.27 seven countries for which this same calculation was As is evident in table 5, as in the 1990s, the expansion performed for a period in the 1990s (Weller, 2000, p. 157), of wage employment for highly skilled workers was shows slightly higher labour demand for less educated centred in the tertiary sector. Services —the sum of workers, whose numbers fall in unwaged employment community, social and personal services and financial, in the median for these seven countries, but rise in the real estate and business services— accounted for over half wage worker category. This appears to point to a labour of these new jobs. By contrast with the 1990s, changes demand that is more balanced by level of education, in commerce (for example, the large-scale expansion with less bias towards the most skilled. However, in of retail commerce: supermarkets and hypermarkets, the median of 15 countries, both wage and unwaged employment expand for individuals with a low level of schooling, albeit slightly more for unwaged. In any case, 27 The breakdown is as follows: n n compared with the results for the 1990s, when the ratio DDSA= SS+ D A //i ==11ii i ii of job creation for the least skilled in wage employment for i = 1, …, n branches of activity, where: with respect to unwaged employment was around 0.5 S = proportion of workers with a specific level of education in total (for seven countries) in the 2000s labour demand for wage employment. this group was slightly stronger (with a ratio of wage Si = proportion of workers with a specific level of education in the branch of activity i. to unwaged employment of 0.8), as it was as well for Ai = wage employment in branch i as a proportion of total wage the larger group of countries. employment. The bars show the averages corresponding to the start and end years. The first term on the right of the function thus captures the contributions of changes between branches of activity (i.e. different with respect 26 The results are also presented for a group of seven countries for to the rise in wage employment), while the second term captures the which this same information is available for the 1990s (Weller, 2000). contribution of changes within the branch of activity (i.e. variations Notably, those data cover a period prior to the “lost half-decade” that in the share of personnel with a specific level of education in wage began in the late 1990s. employment in that branch of activity).

TABLE 4 Latin America (median for 7 and 15 countries): composition of net additional employment, by level of education, for wage and non-wage workers, around 2002-2012 Level of schooling Group of countries Employment category Up to 9 years 10-12 years 13 years or more Total Employed 3.4 46.7 44.2 100.0 7 countries Waged 3.3 24.7 32.6 82.4 Non waged -5.6 13.8 7.4 17.6 Employed 4.7 46.3 44.2 100.0 15 countries Waged 3.3 26.4 32.6 77.9 Non waged 4.1 13.8 8.6 22.1

Source: prepared by the author, on the basis of data from household surveys conducted in the respective countries. Note: the seven countries presented separately are Argentina, Brazil, Chile, Colombia, Costa Rica, Peru and the Plurinational State of Bolivia. The group of 15 countries also includes Bolivarian Republic of Venezuela, Dominican Republic, Ecuador, El Salvador, Honduras, Mexico, Panama and Paraguay.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 23 malls and so on) pushed up demand for highly skilled Overall, the rise in the proportion of highly skilled personnel, and not only because of the expansion of workers within wage earners was due almost entirely to this area of activity (the effect of intersectoral changes), changes within branches of activity.28 In several branches, but also because of intrasectoral changes. Conversely, especially in the tertiary sector but also in construction, manufacturing made only a weak contribution to demand new jobs were generated on the basis of intersectoral for highly skilled workers, mainly because of the reduction changes, but those additional jobs were largely offset in the sector’s share in total employment. Intrasectoral by losses in other branches of activity. changes did help to increase demand for skilled workers in manufacturing, although —as shown in the last column of table 5 (a)— the share of highly skilled personnel rose less than in other sectors. In relative terms, what stands 28 This does not necessarily mean production shifts; change within a out is the upgrading of skills demanded in construction, branch of activity often consists of the substitution of older workers who are retiring with younger workers who have a higher level of mining and, again, commerce. formal education.

TABLE 5 Latin America (median of 15 countries): contribution of intra- and intersectoral changes to variation in the share of education groups in wage employment, around 2002-2012 (Percentages) (a) Wage-earners with a high level of education (13 years or more) Intrasectoral changes Intersectoral changes Total Relative increase a Total 4.09 0.20 4.29 1.00 Agriculture 0.14 -0.48 -0.34 1.01 Mining 0.07 0.11 0.18 1.19 Manufacturing 0.42 -0.29 0.13 0.72 Construction 0.08 0.22 0.31 1.42 Commerce 0.75 0.27 1.02 1.18 Transport and communications 0.26 0.16 0.42 1.14 Services 2.37 0.22 2.59 0.94

(b) Wage-earners with an intermediate level of education (10-12 years) Intrasectoral changes Intersectoral changes Total Total 3.56 0.84 4.38 Agriculture 0.41 -0.03 0.37 Mining -0.02 0.09 0.07 Manufacturing 1.18 -0.69 0.49 Construction 0.54 0.33 0.88 Commerce 0.80 0.89 1.69 Transport and communications 0.43 0.19 0.62 Services 0.20 0.06 0.25

(c) Wage-earners with a low level of education (up to 9 years) Intrasectoral changes Intersectoral changes Total Total -7.56 -1.01 -8.58 Agriculture -0.61 -1.74 -2.34 Mining -0.06 0.08 0.02 Manufacturing -1.57 -1.24 -2.81 Construction -0.66 0.77 0.11 Commerce -1.50 0.84 -0.66 Transport and communications -0.67 0.24 -0.43 Services -2.49 0.03 -2.46

Source: prepared by the author, on the basis of data from household surveys conducted in the respective countries. a Ratio between percentage growth of the education group in the particular branch of activity, and the percentage growth of the education group across all wage-earners. Note: The countries covered are: Argentina, Bolivarian Republic of Venezuela, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Honduras, Mexico, Panama, Paraguay, Peru and the Plurinational State of Bolivia.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 24 CEPAL REVIEW 114 • DECEMBER 2014

The rise in the proportion of workers with an relatively less skilled workers (at least in terms of income) intermediate level of formal education was also due and helped to close wage gaps. mainly to intrasectoral changes (and, thus, part of the The shift in employment towards medium- and generalized rise in education level). However, intersectoral high-productivity segments also benefited average changes had a much stronger impact for this group than income in the low-productivity segments. As noted for more highly educated wage earners. In this respect, earlier, employment in these segments is heterogeneous. labour demand was notably strong from commerce and, In countries with lower per capita gdp and thus a smaller to a much lesser degree, construction, transport and proportion of wage-earners among the employed, a communications. Commerce also stands out for a large large share of own-account employment reflects supply rise in workers with intermediate levels of education pressure (Weller and Kaldewei, 2013, pp. 35-39). In driven by intrasectoral changes, which underscores these countries, per the theoretical arguments outlined the impact of upgrading as noted earlier in relation to in section IV, average labour income from own-account more skilled workers. Manufacturing also shows clear work is lower than in countries with a smaller proportion signs of upgrading of personnel structure, with sharp of own-account workers (where many of them choose increases in workers with intermediate and (as already to work independently in search of advantages) (Weller, seen) high levels of education. 2012, p. 35). The contraction of employment of workers with Accordingly, a relative reduction in employment low levels of formal education was also concentrated in low-productivity segments should also help to drive in changes within branches of activity, pushed along intrasectoral changes by lessening supply pressure and, in part by upgrading of the educational structure in specifically, should increase average income in these services, manufacturing and commerce. However, segments. Between 1990 and 2000-2002, the average intersectoral shifts also played a role in this process, income of wage workers (excluding professional and especially a reduction in employment in agriculture technical workers) in firms with five or more employees and manufacturing. The expansion of commerce and, in remained constant in poverty line terms, while the income particular, construction, also produced jobs for workers of own-account workers (again excluding professional with a low level of formal education. and technical workers) dropped by 0.6 of a poverty The concentration of economic growth in the tertiary line. By contrast, between 2000-2002 and 2009-2010 sector in 2003-2012 did not, then, reduce demand for the income of both groups rose (on average) by 0.4 and highly qualified workers. As in the 1990s, such demand 0.6 of a poverty line, respectively.31 was in fact concentrated in the tertiary sector (Weller, In sum, better economic conditions favoured the 2000). In commerce, specifically, there are signs of income of those employed at the base of the labour upgrading driving demand for skilled workers.29 As income scale, mainly because of two related mechanisms: well, and unlike in previous years, in some cases shifts stronger labour demand (especially in tertiary sector within —and sometimes between— sectors increased activities) for workers with low and intermediate levels demand for workers with intermediate education and, of education, and declining supply pressure on low- to a lesser extent, with little formal education. Again, productivity segments. these processes occurred especially in the tertiary sector, although internal shifts also favoured expansion of 2. Labour institutions and employment for the group with intermediate education employment characteristics in manufacturing and agriculture.30 This partial change in labour demand patterns appears to have benefited Apart from economic and production context, labour institutions have the greatest impact on the evolution and characteristics of employment. Recent changes in labour institutions have helped bring about improvements, 29 This result contradicts the finding of the World Bank (2012) that the strengthening of non-tradable sectors led to a fall in demand for especially in terms employment quality wage gaps. skilled labour. In the first decade of this century the prevailing 30 Klasen, Otter and Villalobos Barría (2012) found that the natural views on labour institutions shifted in many countries. resources boom in the mid-2000s had a beneficial distributive effect The reforms of the 1990s emphasized expanding the in rural areas in Honduras, by improving the income of agricultural workers. There is also anecdotal evidence from several countries to the effect that reduced labour supply has led to income gains in the agricultural sector. 31 Author’s calculations on the basis of eclac (2012).

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 25 range of hiring modalities, especially by using fixed-term conditions for domestic workers were brought into line contracts instead of the traditional indefinite contract, with those for the general workforce. extending the use of trial periods and facilitating Several countries addressed the gap between outsourcing mechanisms. Reforms also reduced the legislation in place and actual compliance by boosting cost of dismissal, mainly by broadening the definition of resources for labour inspection. Many also developed fair dismissal and introducing unemployment insurance schemes to encourage business and labour formalization, systems based on individual accounts (Vega Ruiz, 2005). specifically in microenterprises and smes (ilo, 2014). Generally speaking, and often as part of Since informality is determined both by aspects of macroeconomic stabilization packages, real minimum the production structure and by institutional factors, wages fell sharply in the 1980s, then stagnated in the formalization policies also help to reduce labour market 1990s. In many cases labour inspection practices also segmentation. In particular, formalization of contractual weakened, amid broader strategies to reduce the role relations is crucial for employment quality, because the and size of the State. As a result, in many cases the gap evidence shows that a formal contract of employment is between legal provisions and the reality at work tended often the key that unlocks other benefits that come with to widen (Bensusán, 2006). good-quality jobs (Weller and Roethlisberger, 2011; At the same time, the proportion of employment eclac/ilo/fao, 2012). in medium- and high-productivity segments fell (owing A number of countries adopted reforms with a to low rates of economic growth, a sharp fall in the protective slant regarding individual labour relations, for proportion of public-sector employment and, at least example by reducing weekly working hours, increasing temporarily, a drop in wage employment-gdp elasticity), compensation for unfair dismissal, restricting the use which led to “de facto flexibilization”. It also sapped of overtime or increasing overtime pay, and extending the power of the trade unions. prenatal and postnatal leave. The poor results of the reforms of the 1980s At the same some, some countries strengthened and 1990s and the various crises that broke out in unemployment insurance schemes or created new Latin America and the Caribbean after the mid-1990s models (and a number of countries are in the process of undermined the credibility of deregulation-based labour doing so) to broaden systems to protect workers in the policy as a tool for job creation. Against this backdrop, in context of more volatile markets. Many countries have the 2000s new political proposals emerged in the region taken a more active stance on minimum wage policy, under the premise that the response to globalization should as well.33 As in the 1990s, in Mexico the drop in the not be a one-fits-all approach to production restructuring minimum wage was identified as having contributed and economic and labour policy (Fraile, 2009; Weller, significantly to broadening income gaps (Cortez, 2001; 2009; Lee and McCann, 2011). At the same time, radical Bosch and Manacorda, 2010). Concerning the recent deregulation began to be treated with increasing caution period, Cornia (2014), Keifman and Maurizio (2012), in the international debate (idb, 2003; Freeman, 2005). and eclac (2014c) all found that minimum wage In Latin America, scepticism regarding deregulation gains and increased formality —or the combination of is also driven by the fact that large-scale liberalizing both—have been important factors in narrowing in gaps reforms were considered to have had little impact on in several countries. economic growth (idb, 1997), and that the expectations In short, unlike during preceding decades, in the of formal jobs creation attached to the reforms were recent period stronger measures were taken to strengthen not fulfilled. workers’ individual and collective rights in respect of In the 2000s, trade union organization improved in the twofold purpose of labour institutions mentioned some countries and the decline in organized labour eased earlier, supported by measures such as business and or reversed.32 In this scenario, collective bargaining began labour formalization policies, albeit generally without to cover a slightly wider range of issues and new groups leaving aside the aim of labour market efficiency. of workers, such as seasonal and domestic workers. In some cases, outsourcing was regulated after evidence that this tool tended to be abused, and employment 33 In the median for 20 countries, the real minimum wage shows no change in 1990-2003, but increased by 2.0% in 2003-2012. If the changes are weighted by the working-age population, the annual 32 See ilo (2009) for an illustration of the decline in labour organization increase for the recent period was 4.4% (author’s calculation on the between 1989 and 2005. basis of official data from the countries).

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 26 CEPAL REVIEW 114 • DECEMBER 2014

V Conclusions

In the period 2003-2012, the performance of the labour large productivity gaps, a high percentage of poor markets of Latin America and the Caribbean represented and unprotected workers, low levels of labour market a departure from previous patterns in several aspects. In participation by women, sharp labour inequalities particular, the open unemployment rate fell noticeably between men and women both within and outside thanks to a rise in employment levels, employment labour markets, discrimination against various groups of quality indicators improved in a context of labour workers, poor employment quality indicators, and low formalization, and income gaps between more and less rates of continuous training, trade union membership skilled workers narrowed. and collective bargaining. Obviously, recent advances To understand the factors influencing this notwithstanding, the challenges remain considerable. performance, this article proposed analysing the behaviour A number of factors threaten to at least slow the pace of the economic and production context (economic growth, of labour improvements in the near future. First, amid labour productivity trends and changes in structural slower global economic growth than in the recent past, heterogeneity) and of labour institutions. Review of the outlook for regional growth is less promising amid these factors for the decade considered shows that the slacker external demand (eclac, 2014b). At the same combination of the two factors shaped the three stylized time, in many countries domestic demand growth (driven facts characterizing recent labour performance. Relatively chiefly by household consumption) is slowing owing to high economic growth stimulated labour demand, which factors such as the end of the terms-of-trade gain (which boosted employment generation, especially in medium- slows the increase in disposable national income), the and high-productivity segments, facilitating increased end of a period of exchange-rate appreciation and high formality and higher labour income. The characteristics levels of household indebtedness. At the regional level, of labour demand, centred on non-tradable sectors (the employment generation was already weak by 2013, both tertiary sector and construction), set the patterns for in absolute terms and in relation to that year’s economic employment generation. Intra- and intersectoral shifts growth (eclac, 2014a). led to rises in productivity, related —in several branches From the policy standpoint, the tendency in many of activity, such as commerce and transport, storage and countries to focus formalization efforts on incentives and communications— to upgrading of the occupational more stringent oversight could start to yield diminishing structure, which helped to improve employment quality. returns. This is because many firms and workers with Labour demand characteristics improved opportunities for the production potential to assume the necessary costs workers with low and intermediate levels of education, as have already done so, while other firms and workers well, which helped to narrow income gaps. Institutional still operating in informal conditions are labour-supply changes also contributed to narrowing these gaps, to driven and their productivity is too low to support employment formalization and to improvements in full formalization. It may, then, be time to rethink employment quality indicators. Strong employment instruments for promoting development, especially of generation was a key factor in bringing down poverty microenterprises and smes. It is also important to strengthen rates, as was the reduction in income gaps for lessening the virtuous cycle between labour productivity gains and inequality at the household level. employment quality. Key steps to this end are improving Despite the progress noted, which reversed previous education, professional training and capacity-building, trends, labour markets in the region continued to show and strengthening the voice of workers by expanding severe weaknesses, such as high levels of informality, organized labour and collective bargaining.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER CEPAL REVIEW 114 • DECEMBER 2014 27

Bibliography Akerlof, George A. and Janet Louise Yellen (1986), Efficiency (2010), Time for equality: closing gaps, opening trails Wage Models of the Labor Market, New York, Cambridge (LC/G.2432(SES.33/3)), Santiago, Chile. University Press. (2007), “Progreso técnico y cambio estructural en Azevedo, João Pedro and others (2013), “Is labor income responsible América Latina”, Project Document, No. 136 (LC/W.136), for poverty reduction? A decomposition approach”, Policy Santiago, Chile. Research Working Papers, No. 6414, Washington, D.C., (various years), Economic Survey of Latin America and the World Bank. Caribbean, Santiago, Chile. Azevedo, João Pedro, Gabriela Inchauste and Viviane Sanfelice eclac/ilo (Economic Commission for Latin America and the (2013), “Decomposing the recent inequality decline in Latin Caribbean/International Labour Organization) (2013), “Youth America”, Policy Research Working Papers, No. 6715, employment: crisis and recovery”, The Employment Situation Washington, D.C., World Bank. in Latin America and the Caribbean, No. 7, Santiago, Chile, Ball, Laurence M., Nicolás de Roux and Marc Hofstetter (2011), October. “Unemployment in Latin America and the Caribbean”, nber (2012), “Labour productivity and distribution issues”, The Working Paper, No. 17274, Cambridge, Massachusetts, Employment Situation in Latin America and the Caribbean, National Bureau of Economic Research. No. 6, Santiago, Chile, May. Bensusán, Graciela (2008), “Regulaciones laborales, calidad de los eclac/ilo/fao (Economic Commission for Latin America and empleos y modelos de inspección: México en el contexto the Caribbean/International Labour Organization/Food and latinoamericano” (LC/MEX/L.861), Mexico City, eclac Agriculture Organization of the United Nations) (2012), Subregional Headquarters in Mexico. Labour Market Policies and Rural Development in Latin (coord.) (2006), Diseño legal y desempeño real: Instituciones America, Food and Agriculture Organization of the United laborales en América Latina, Mexico City, Chamber of Nations (fao), Rome. Deputies/Metropolitan Autonomous University. Eichhorst, Werner, Michael Feil and Christoph Braun (2008), “What Berg, Janine and David Kucera (2008), “Labour institutions in the have we learned? Assessing labor market institutions and developing world: historical and theoretical perspectives”, In indicators”, iza Discussion Papers, No. 3470, Bonn, Institute Defence of Labour Market Institutions. Cultivating Justice in for the Study of Labour (iza). the Developing World, Janine Berg and David Kucera (eds.), Fields, Gary (2004), A Guide to Multisector Labor Market Models, Basingstoke, Palgrave MacMillan. Washington, D.C., November. Berman, Eli, John Bound and Zvi Griliches (1994), “Changes in the Fraile, Lydia (2009), “Lessons from Latin America’s neo-liberal demand for skilled labor within U.S. manfacturing: evidence experiment: An overview of labour and social policies since from the Annual Survey of Manufactures”, The Quarterly the 1980s”, International Labour Review, vol. 128, No. 3. Journal of Economics, vol. 109, No. 2, Oxford University Freeman, Richard B. (2005), “Labor market institutions without Press, May. blinders: the debate over flexibility and labor market Bosch, Mariano, M. Belén Cobacho and Carmen Pagés (2012), performance”, nber Working Paper, No. 11286, Cambridge, “Taking stock of nine years of implementation of Seguro Massachusetts, National Bureau of Economic Research. Popular in Mexico”, Technical Notes, No. IDB-TN-442, Gasparini, Leonardo and others (2011), “Educational upgrading and Washington, D.C., Inter-American Development Bank. returns to skills in Latin America. Evidence from a supply- Bosch, Mariano and Marco Manacorda (2010), “Minimum wages demand framework, 1990-2010”, Policy Research Working and earnings inequality in urban Mexico”, American Economic Paper, No. 5921, Washington, D.C., World Bank. Journal: Applied Economics, vol. 2, No. 4, Nashville, idb (Inter-American Development Bank) (2003), Good Jobs Wanted. Tennessee, American Economic Association. Labor Markets in Latin America, Washington, D.C. Cornia, Giovanni Andrea (2014), “Income inequality in Latin (1997), Economic and Social Progress in Latin America. America. Recent decline and prospects for its further 1997 Report, Washington, D.C. reduction”, Macroeconomía del Desarrollo series, No. 149 ilo (International Labour Organization) (2014), Recent Experiences (LC/L.3847), Santiago, Chile, Economic Commission for Latin of Formalization in Latin America and the Caribbean, Lima, America and the Caribbean (eclac). ilo Regional Office for Latin America and the Caribbean. Cortez, Willy W. (2001), “What is behind increasing wage (2013), “Key Indicators of the Labour Market (kilm)”, inequality in Mexico?”, World Development, vol. 29, No. 11, eigth edition [online] http://www.ilo.org/empelm/what/ Amsterdam, Elsevier. WCMS_114240/lang--en/index.htm. Cruces, Guillermo, Carolina García Domenech and Leonardo (2009), World of Work Report 2008. Income Inequalities Gasparini (2012), “Inequality in education: evidence for Latin in the Age of Financial Globalization, Geneva. America”, Working Paper, No. 135, Center for Distributive, (various years), Panorama Laboral de América Latina y Labor and Social Studies (cedlas). el Caribe, Lima, Regional Office for Latin America and the eclac (Economic Commission for Latin America and the Caribbean) Caribbean. (2014a), Preliminary Overview of the Economies of Latin Infante, Ricardo (ed.) (2011), “El desarrollo inclusivo en América America and the Caribbean, 2013 (LC/G.2581-P), Santiago, Latina y el Caribe. Ensayos sobre políticas de convergencia Chile. United Nations publication, Sales No. E.14.II.G.2. productiva para la igualdad”, Libros de la cepal, No. 112 (2014b), Economic Survey of Latin America and the (LC/G.2500-P), Santiago, Chile, Economic Commission for Caribbean 2014 (LC/G.2619-P), Santiago, Chile. United Latin America and the Caribbean (eclac). United Nations Nations publication, Sales No. E.14.II.G.3. publication, Sales No. S.11.II.G.56. (2014c), Compacts for Equality: Towards a Sustainable Kacef, Osvaldo and Rafael López-Monti (2010), “Latin America, Future (LC/G.2586(SES.35/3)), Santiago, Chile. from boom to crisis: macroeconomic policy challenges”, cepal (2013), Social Panorama of Latin America 2013 Review, No. 100 (LC/G.2442-P), Santiago, Chile. (LC/G.2580), Santiago, Chile. United Nations publication, Keifman, Saúl N. and Roxana Maurizio (2012), “Changes in Sales No. E.14.II.G.6. labour market conditions and policies. Their impact on wage (2012), Social Panorama of Latin America 2011 inequality during the last decade”, wider Working Papers, No. (LC/ G.2514-P), Santiago, Chile. United Nations publication, 2012/14, Helsinki, United Nations University/World Institute Sales No. E.12.II.G.6. for Development Economics Research.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER 28 CEPAL REVIEW 114 • DECEMBER 2014

Klasen, Stephan, Thomas Otter and Carlos Villalobos Barría (2012), Employment Programme for Latin America and the “The dynamics of inequality change in a highly dualistic Caribbean (prealc). economy: Honduras, 1991-2007”, wider Working Papers, Vega Ruiz, María Luz (2005), La reforma laboral en América Latina: No. 2012/17, Helsinki, United Nations University/World 15 años después, Lima, ilo Regional Office for Latin America Institute for Development Economics Research. and the Caribbean. Lee, Sangheon and Deirdre McCann (2011), Regulating for Decent Weller, Jürgen (2012), “Crecimiento, empleo y distribución Work: New Directions in Labour Market Regulation, Geneva, de ingresos en América Latina”, Macroeconomía del International Labour Organization (ilo). Desarrollo series, No. 122 (LC/L.3516), Santiago, López-Calva, Luis F. and Nora Lustig (eds.) (2010), Declining Chile, Economic Commission for Latin America and the Inequality in Latin America. A Decade of Progress?, Caribbean (eclac). Washington, D.C., Brookings Institution Press. (2009), “Avances y retos para el perfeccionamiento de McMillan, Margaret S. and Dani Rodrik (2011), “Globalization, la institucionalidad laboral en América Latina”, El nuevo structural change and productivity growth”, nber Working escenario laboral latinoamericano. Regulación, protección y Paper, No. 17143, Cambridge, Massachusetts, National Bureau políticas activas en los mercados de trabajo, Jürgen Weller of Economic Research. (ed.), Buenos Aires, Siglo XXI Editores. Ocampo, José Antonio, Codrina Rada and Lance Taylor (2009), (2000), Reformas económicas, crecimiento y empleo: Los Growth and Policy in Developing Countries: A Structuralist mercados de trabajo en América Latina, Santiago de Chile, Approach, New York, Columbia University Press. Economic Commission for Latin America and the Caribbean Pagés, Carmen, Gaëlle Pierre and Stefano Scarpetta (2009), Job (eclac)/Fondo de Cultura Económica. Creation in Latin America and the Caribbean. Recent Trends Weller, Jürgen and Cornelia Kaldewei (2013), “Empleo, crecimiento and Policy Challenges, Washington, D.C., World Bank. sostenible e igualdad”, Macroeconomía del Desarrollo series, Perry, Guillermo E. and others (2007), Informality: Exit and No. 145 (LC/L.3743), Santiago, Chile, Economic Commission Exclusion, Washington, D.C., World Bank. for Latin America and the Caribbean (eclac). Rodrik, Dani (2013), “The past, present, and future of economic Weller, Jürgen and Claudia Roethlisberger (2011), “La calidad del growth”, Working Paper, No. 1, Global Citizen Foundation, June. empleo en América Latina”, Macroeconomía del Desarrollo Ros, Jaime (2011), “La productividad y el desarrollo en América series, No. 110 (LC/L.3320-P), Santiago de Chile, Economic Latina: Dos interpretaciones”, Economía unam, vol. 8, No. 23, Commission for Latin America and the Caribbean (eclac). Mexico City, National Autonomous University of Mexico. United Nations publication, Sales No. S.11.II.G.39. Tokman, Víctor (1987), “El sector informal: Quince años después”, World Bank (2012), The Labor Market Story Behind Latin America’s Documento de Trabajo, No. 316, Santiago, Chile, Regional Transformation, Washington, D.C.

ASPECTS OF RECENT DEVELOPMENTS IN THE LATIN AMERICAN AND CARIBBEAN LABOUR MARKETS • JÜRGEN WELLER The earnings share of total income in Latin America, 1990-2010

Martín Abeles, Verónica Amarante and Daniel Vega

ABSTRACT This article analyses the share of total income represented by employment earnings in the countries of Latin America over the last two decades. It first considers the wage

share of gross domestic product (gdp) and then adds in the earnings of self-employed workers. The findings indicate that both total wages and total earnings declined as a

share of gdp in most of the region’s countries over the period, although there were some exceptions. The reduction in earnings inequality seen over the past decade was

not usually accompanied by an increase in the gdp share of earnings. This means that the improvement in personal income distribution was not matched by an improvement in functional distribution.

KEYWORDS Employment, income, gross domestic product, income distribution, measurement, statistical data, Latin America

JEL CLASSIFICATION D33, E25, J31

AUTHORS Martín Abeles is the director of the eclac office in Buenos Aires. [email protected] Verónica Amarante is the director of the eclac office in Montevideo. [email protected] Daniel Vega is a statistical assistant with the eclac office in Buenos Aires. [email protected] 30 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

In recent decades, the great majority of studies on income of self-employed workers (both own-account inequality have concentrated on analysing income workers and employers) comes from earnings. Once distribution among individuals and households. Although the self-employment earnings total has been identified, functional income distribution has received greater the proportion it bears to the wage total is calculated, attention in advanced countries over recent years, largely once again from household survey information. On the because of the decline in the wage share of income in basis of this ratio, the wage total identified by systems those countries during the past three decades, the subject of national accounts is adjusted to obtain an estimate of does not seem to have attracted the same interest in the total earnings as a share of gdp. On the basis of these countries of the Latin America region. This article argues two methodological approaches, the article presents new that functional income distribution needs to be studied estimates for the volume of earnings in Latin America, systematically and discusses some of the information while also illustrating their evolution and reflecting constraints that usually hinder efforts to do so, particularly on the problems and constraints involved in studying in the countries of Latin America. functional income distribution in the region. The recent trend of earnings as a share of total gdp in The article is organized as follows. First, the the region’s economies is analysed. The empirical evidence functional and personal approaches to studying income presented is structured around two main categories of distribution are discussed (section II). Some issues involved earnings. First, information on the total wage share in in measuring the earnings share of total income are then the region’s countries during the period from 1990 to described (section III), after which recent studies on 2011 is considered. Second, the wage share of gdp is functional income distribution are summarized (section IV). corrected by adding self-employment earnings. Two The following section discusses methodological methods are used to make this adjustment. The first is aspects, presenting the information sources available for to assume, as is very commonly done in the literature, studies of this type in Latin America and describing the that all self-employed workers earn the average wage methodological options adopted in this paper (section V). in the economy. The second is more rigorous and uses Thereafter, the evolution of total wages as a share of household survey data. The first step in this second total income in Latin America is analysed (section VI). adjustment is to use information from continuous Having highlighted the importance of self-employment household surveys to estimate what portion of the mixed work in Latin America (section VII), the article presents the results arrived at when this wage total is corrected to incorporate the earnings of self-employed workers The authors are grateful to Maximiliano García and Martin Schmitt (section VIII). Lastly, section IX contains some for their outstanding work as research assistants on this study. concluding remarks.

II Functional and personal income distribution

The classical economists were concerned to analyse landowners was the chief problem of political economy the relationship between functional income distribution (Ricardo, 1973). As Serrano and Medeiros (2001) argue, and the production and capital formation process. In the idea of an economic surplus as conceived in the the now famous commentary at the start of his On the classical approach permeated to varying degrees into a Principles of Political Economy and Taxation, David number of the seminal works on economic development, Ricardo argued that determining the laws that governed as in the case of the well-known dual economy model the division of output between workers, capitalists and propounded by Arthur Lewis (1954). Analysis of the

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 31 relationship between economic development and falling the consolidation of welfare States has given rise to employment in traditional or subsistence activities, a need for more accurate identification of the most typically in rural areas (see Bhaduri, 1983), and indeed vulnerable social groups, the main beneficiaries of of the relationship between the creation of the surplus public assistance, and for a more rigorous examination and capital accumulation from a sectoral and structural of income distribution within the working class. The perspective (Rodríguez, 2006), entails some kind of personal approach has made it possible to analyse the analysis of the distribution of the surplus between the impact of redistributive State action more thoroughly different social classes or sectors. by using statistical information from household surveys The classical emphasis on functional income to consider the effects of taxes and transfers on income. distribution survived the marginal revolution of the late Thus, the growing popularity of the personal approach nineteenth century, albeit within a different conceptual was also connected with the increasing use of specific and methodological framework whereby each factor of surveys to collect household-level information and production (the social classes of the classical authors) with methodological and technological advances in appropriated a portion of output on the basis of its marginal applied research. contribution to the production process. A fundamental It might also be said that analysis of the evolution indicator in this school of thought is the wage share of functional income distribution tended to fall out of of total output in the economy. The consolidation of favour because the data on the wage share of income neoclassical economics around the mid-1950s, with were so stable in the period after the Second World War, its stress on analysis based on the study of economic and this empirical observation came to be identified agents’ individual behaviour, led to the emphasis among as a stylized fact of capitalist economies (Kaldor, mainstream economists shifting from functional income 1961). This empirical regularity would be reinforced distribution to personal income distribution (Goldfarb and in turn by the consolidation of neoclassical growth Leonard, 2005). At the outset, this interest in studying theory and the idea that the production potential of personal income distribution as opposed to functional an economy could be captured by a Cobb-Douglas distribution was resisted by post-Keynesian and neo- production function,3 which implies an elasticity of Ricardian economists, who argued for the primacy of substitution of 1 between labour and capital, and constant the functional distribution debate not only on analytical factor shares. grounds1 but with the explicit objective of stressing the The subject has come back to prominence in recent centrality of the social conflict between capitalists and years, however. In academia, a number of authors working workers in the development of capitalist economies.2 in the post-Keynesian or structuralist tradition, or both, They thus sought to highlight the idea that individual have systematically studied the impact of changes in choices were heavily influenced by each person’s position functional income distribution on aggregate demand in the social stratification. The shift in emphasis towards and economic growth.4 Empirically, the assumed personal distribution became patent in the 1960s and stability of functional income distribution was called analysis of functional income distribution was relegated into question by the downward trend observed in the to the background (Atkinson, 2009). wage share of income in virtually all industrialized Different historical or institutional factors drove countries from the early 1980s (see section III below). the growing interest in personal income distribution. This new development even aroused the attention of The complexity of modern production processes and academia, as a number of recent studies show.5 Greater the internal heterogeneity of the groups associated with the different factors of production help to explain why analyses of inequality have tended to centre on personal 3 According to Paul Douglas, the division of national income between distribution. Not only can individuals and households capital and labour was roughly constant for a long period. The most derive income from different factors of production, but recent United States data are also consistent with a Cobb-Douglas production function. Despite the various changes in the economy inequality can be very high within a given group (wage over the past four decades, the division of income is easily explained earners, for example). From an institutional perspective, by a Cobb-Douglas production function (Mankiw, 2007, pp. 55-58, cited in Atkinson, 2009). 4 See, for example, Bhaduri and Marglin (1990) and Taylor (1991) on the subject of seminal contributions. For a summary of this literature, 1 Among other things, they pointed out the need for macroeconomic see Abeles and Toledo (2011). analysis to differentiate the various social classes’ propensity to save. 5 See, for example, Bentolila and Saint-Paul (2003); Gollin (2002); 2 Authors such as Joan Robinson, Nicholas Kaldor and Luigi Pasinetti Serres, Scarpetta and Maisonneuve (2001); Feldstein (2008); imf (2007), developed arguments along these lines. European Commission (2007); ilo (2008) and Ellis and Smith (2010).

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 32 CEPAL REVIEW 114 • DECEMBER 2014 interest was also shown in the subject in developing development process (eclac, 2012a), especially in the countries. In Latin America and the Caribbean, for South American countries, and estimates of the rents example, the export commodity price boom gave a associated with natural resource extraction have become renewed centrality to the implications of the scale an important policymaking input (Campodónico, 2008; and sectoral origin of the economic surplus for the eclac, 2013).

III Measuring the earnings share of total income

The nature of the process whereby income is generated All that is usually considered when the earnings from economic activity is reflected in the income share of total income is analysed is the ratio of wages to generation account of the System of National Accounts total gross domestic product (gdp).8 This ratio may be (sna). This account shows how gross value added is expressed at market prices or at factor cost, depending on distributed between workers, the owners of capital whether taxes on products and production less subsidies (including land and other natural resources whose are included in the gdp measurement. Taken alone, rents can be appropriated by private individuals) and this ratio shows that the wage share varies enormously the government. In the 1993 revision, value added is between countries, ranging from 14% in Nigeria to treated as a resource in the income generation process, 59% in Switzerland, and the region’s countries tends while remuneration for wage employees and taxes on to present low values for this indicator (see figure 1). products and production less subsidies are treated as One regularity that emerges when the wage share use. The accounting balance is the operating surplus or of gdp around the world is analysed is the positive mixed income, depending on the nature of the account. association with a country’s level of wealth. Wages Mixed income reflects the surplus yielded by the represent a larger share of gdp in richer countries (see production activities of unincorporated businesses, i.e., figure 2). This relationship may be deceptive, however, households.6 It thus implicitly contains an element of since the fact that the share rises with per capita gdp remuneration for both the labour and the capital involved may be due to the greater size of the informal economy in the production activity being analysed. At best, the in less developed countries. Thus, there may be large national accounts provide aggregate information on biases in comparisons between different countries and these two components of mixed income, and cannot time periods. Comparisons across time will be affected separately identify the yield of labour and capital.7 This if, as the evidence suggests, wage incomes and self- is one of the main constraints on efforts to measure the employment incomes react differently to the economic wage share of income, most particularly in developing cycle. Comparisons between countries, meanwhile, can be countries, where self-employment prevails. The total expected to be heavily skewed by non-inclusion of self- income of self-employed workers is usually included in employment income, as this will result in underestimation this mixed income, with no distinction made between of the total income share going to labour: the greater the the remuneration of labour and capital. proportion of self-employment and thence the overall earnings of these excluded workers, the greater the underestimation will be.

6 These are essentially self-employed or own-account workers (street vendors, for example). 7 In some of the region’s countries, such as Peru between 1950 and 1965, self-employed workers’ income used to be published as part of 8 Labour force participation studies based on data from surveys of the national accounts statistics, but the information was discontinued businesses or industrial censuses have a similar limitation, as they as national practices were adapted to United Nations proposals for measure the activity of firms above a certain size threshold (whether of standardizing systems of national accounts. See [online] http:// output or number of employees), so that small or family businesses are institutodelperu.org.pe/descargas/Publicaciones/De%20otras%20 excluded (see Rodríguez and Ortega, 2006, for a detailed description entidades/DOC/1966_webb_cuentas_nacionales_del_peru.pdf. of the databases in which this type of information is compiled).

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 33

FIGURE 1 Wages as a share of gdp at market prices, around 2010 (Percentages) 70

60

50

40

30

20

10

0 Italy India Chile Japan Spain Niger Malta Brazil Latvia Serbia France Poland Greece Ireland Guinea Cyprus Austria Kuwait Canada Mexico Belarus Finland Panama Senegal Sweden Ukraine Norway Portugal Bulgaria Belgium Uruguay Hungary Slovenia Slovakia Denmark Germany Lithuania Mongolia Sri Lanka Colombia Argentina Guatemala Costa Rica Azerbaijan Kyrgyzstan Kazakhstan Switzerland Netherlands South Africa Mozambique United States Macao (China) Czech Republic United Kingdom Republic of Korea Russian Federation Hong Kong (China) Hong Kong Republic of Moldova Plurinational State of Bolivia Bolivarian Republic of Venezuela Former Yugoslav Republic of Macedonia Source: prepared by the authors, on the basis of data from the United Nations Statistics Division (unsd). gdp: gross domestic product.

FIGURE 2 Wages as a share of gdp and per capita gdp at market prices, 2010 (Percentages) 70

60 CHE

GBR DNK USA SVN FRA SWE LKA CAN JPN FINDEU BEL NLD HKG 50 UKR CRI BLR RUS PRT AUT ESP CYP NOR ZAF KOR IRL MDA BRASRB LVA HUN MLT CZE ITA 40 LTU MKD SVK BGR POL ARGCHL GRC KAZURY COL PAN GTM 30 KGZIND VEN MEX BOL MOZ MNG KWT MAC

20 SEN

Wages as a share of  (percentages) AZE NERGIN 10 y = 0.0696ln(x) - 0.2643 R2 = 0.4212

0 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 Per capita  at purchasing power parity (2005 dollars) Source: prepared by the authors, on the basis of data from the United Nations Statistics Division (unsd) and International Monetary Fund (imf), World Economic Outlook Database. gdp: gross domestic product. che: Switzerland, dnk: Denmark, usa: United States, gbr: United Kingdom, svn: Slovenia, fra: France, swe: Sweden, can: Canada, fin: Finland, bel: Belgium, lka: Sri Lanka, jpn: Japan, deu: Germany, hkg: Hong Kong (China), nld: Netherlands, ukr: Ukraine, cri: Costa Rica, blr: Belarus, rus: Russian Federation, prt: Portugal, esp: Spain, aut: Austria, mda: Republic of Moldova, zaf: South Africa, bra: Brazil, srb: Serbia and Montenegro, lva: Latvia, hun: Hungary, kor: Republic of Korea, cyp: Cyprus, irl: Ireland, cze: Czech Republic, ita: Italy, ltu: Lithuania, nor: Norway, mlt: Malta, mkd: Former Yugoslav Republic of Macedonia, svk: Slovakia, pol: Poland, bgr: Bulgaria, grc: Greece, chl: Chile, arg: Argentina, ury: Uruguay, pan: Panama, col: Colombia, ven: Bolivarian Republic of Venezuela, mex: Mexico, gtm: Guatemala, bol: Plurinational State of Bolivia, kwt: Kuwait, moz: Mozambique, mac: Macao (China), mng: Mongolia, sen: Senegal, aze: Azerbaijan, gin: Guinea, ner: Nigeria.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 34 CEPAL REVIEW 114 • DECEMBER 2014

IV Recent studies on the earnings share of total income and its determinants

Recent studies of functional income distribution have processes, labour market deregulation, the deregulation all identified a significant change in the last three and increasing predominance of financial markets, and decades. By contrast with the so-called trente glorieuses, changes in institutions and the degree of unionization, the three decades of strong growth that followed the among other things (Bentolila and Saint-Paul, 2003; Second World War, the earnings share of total income Bernanke and Gürkaynak, 2002; Fichtenbaum, 2009; has been falling in the countries of the Organisation Gollin, 2002; Harrison, 2002; Hogrefe and Kappler, for Economic Co-operation and Development. The 2012; imf, 2007; Jayadev, 2007; Rodríguez and Ortega, trend is less homogeneous in developing countries and 2006). However, there has been little in the way of emerging economies, although the earnings share has systematic approaches to the link between the evolution also been dropping in most of them. Recent studies of functional and personal income distribution. While providing evidence of this are Stockhammer (2013) and studies of functional distribution take a macroeconomic ilo (2011 and 2013). These studies look at the evolution approach, those centring on personal distribution treat of the adjusted wage total, calculated by multiplying the it as a microeconomic phenomenon, accounted for average compensation of wage earners by the number basically by the distribution of individuals’ personal of workers in the economy. Self-employed workers characteristics. One attempt to relate the two approaches are thus incorporated on the assumption that they earn can be found in a study by Daudey and García-Peñalosa roughly the same as wage workers. (2007) providing econometric evidence that a low wage The drop in the earnings share of total income has share of output has a negative and significant effect on not generally been due to structural changes in economic personal income inequality. activity involving a shift from sectors with a large wage For a better understanding of the determinants share to others with a smaller share (displacement effects). of income inequality and the connection between Rather, it has been due to a decline in the wage share personal and functional inequality, it is first necessary within certain sectors (ilo, 2011). In particular, the ratio to have an accurate diagnosis of functional income has declined considerably in financial intermediation distribution and its evolution. There are analyses from and high- and medium-technology manufacturing, with a factorial perspective for some countries in the region, a less pronounced drop in services, construction and including Lindenboim (2008), Lindenboim, Kennedy low-technology manufacturing. It has also been found and Graña (2010) and Graña (2007) for Argentina; that the downward trend in the adjusted earnings share in Hernández Laos (1998) for Mexico; undp (2010) developed countries has been mainly due to the declining and Amarante and Vigorito (2011) for Uruguay; and share of low- and medium-skilled workers’ earnings, Ministry of Planning and Cooperation (2000) for while the earnings share of highly skilled workers has Chile. However, there is no analysis combining a tended to rise (ilo, 2013). comparative perspective in the region with systematic, In recent years, a quite substantial literature has comparable inclusion of self-employment income. The attempted to relate the recent evolution of the earnings present article seeks to advance in this area, and while share with the structural reforms implemented in the the matter may seem straightforward, a number of past few decades, analysing possible links with various information availability problems have to be dealt with, as developments such as global offshoring of production detailed below.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 35

V Methodology

1. The information available in Latin America of the different series by business sector.12 However, the data cover a more limited time period, with information National accounts statistics in the region are prepared going back to the 1970s for just 7 of the 16 countries. by central banks or national institutes of statistics.9 Not The detail of the information on Latin America held by all countries make disaggregated information on mixed unsd is presented in table A.2 of the annex. income publicly available. In the region, this information This paper uses information from the cepalstat is available for Argentina, the Bolivarian Republic database, supplemented by information from the relevant of Venezuela, Brazil, Chile, Colombia, Guatemala, official bodies for countries not included in cepalstat Honduras, Mexico, Nicaragua, Panama, Paraguay, (Argentina, Costa Rica and Guatemala). In the case of Peru and Uruguay, although in most cases only for the Uruguay, where sna information only runs up to 2005, most recent period (usually since the last base change), the wage ratio was updated in line with the average while in some cases it is not disaggregated by branch of nominal wage index, the employment rate and gdp at activity.10 This information is assembled by the databases factor cost. of the Economic Commission for Latin America and the Caribbean (eclac) and the United Nations Statistics 2. Methodologies for estimating total earnings Division (unsd).11 The eclac data are available in cepalstat, a To avoid the biases in the measurement of the earnings database that includes information for recent years on share that arise when the analysis does not cover the 14 of the region’s 18 countries (not counting Cuba or totality of earnings but only wage income, as discussed Haiti), without distinguishing mixed income, which is in section III, the earnings of self-employed workers consolidated with the operating surplus (even in cases need to be estimated. where the information available from the countries is A first problem to be addressed, then, is how to disaggregated). The detail of the information available separate out from the mixed income total the share in this database is presented in table A.1 of the annex. deriving from the remuneration of labour and the share One advantage of the database is that it holds long- deriving from returns to capital. The first component term information, although in practice there are large should be added to wage employees’ remuneration to hiatuses in the time series because of base changes or obtain the true labour share of income generated in the other alterations in methodology. economy. Different methods have been suggested for unsd holds information on a great many countries, making this correction. One possibility is to carry out including 16 in the region, distinguishing mixed income estimates on the assumption that self-employed workers from the operating surplus and including a breakdown earn roughly the average wage. Gollin (2002) makes adjustments of this type for a large group of countries, and concludes that a substantial part of the differences in the earnings share of total income between rich and poor 9 Institutes of statistics are responsible for gathering this information countries is due to methodological errors caused by the in Argentina, Brazil, Colombia, Mexico, Panama, Peru and the non-inclusion of self-employment incomes. Studies by Plurinational State of Bolivia, while central banks do so in the Bolivarian Republic of Venezuela, Chile, Costa Rica, the Dominican Republic, Stockhammer (2013) and ilo (2011 and 2013) also rely Ecuador, El Salvador, Guatemala, Honduras, Nicaragua, Paraguay on a correction of this type. The present paper carries out and Uruguay. estimates using two methodologies, the first of which 10 Some countries have information on mixed income prior to 1990 consists in assuming that self-employed workers earn in their systems of national accounts. 11 There are other sources of information for analysing the wage share roughly as much on average as wage employees. of gdp, based not on national accounts data, however, but on business surveys or censuses. The two main databases of this type are those of the United Nations Industrial Development Organization (unido) and the Organisation for Economic Co-operation and Development 12 This total does not include Ecuador, whose data only go up to (oecd). Details of these databases can be found in Rodríguez and 1991. In the breakdown by business sector, mixed income figures Ortega (2006). are not always available.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 36 CEPAL REVIEW 114 • DECEMBER 2014

A more rigorous solution is to try to separate As a first step, the employment income of self- the employment remuneration share from capital employed workers is estimated in the light of their personal returns in the mixed income total for countries that characteristics and economic sector, on the assumption calculate it separately. For this, the earnings of self- that their earnings are similar to those of wage employees. employed workers can be simulated by considering The procedure is to start by estimating wage equations their personal characteristics and the economic sector for private-sector wage earners, using as dependent they work in (rather than using the average wage for all variables sex, age and age squared, years of education and of them). A solution of this type is adopted by Young binary variables distinguishing by branch of activity. The (1995) for the countries of Asia on the basis of census coefficients obtained in these wage equations are applied information, on the assumption that self-employed to the characteristics of self-employed workers so that workers earn the same as wage workers who are similar the earnings of each self-employed worker included in in age, sex, education level and economic sector. Even the survey can be predicted. When these predictions for when no official information is available on mixed the earnings of self-employed workers are lower than the income, the estimate of the earnings share can be income reported in household surveys, the difference is corrected. To this end, the household survey is assumed to represent returns to capital. Accordingly, the first used to calculate the ratio between the wage figure yielded by the prediction is taken to be the amount total and self-employment income, and then this earned by these self-employed workers. If the earnings coefficient is applied to the wage income recorded in the predicted for self-employed workers are higher than national accounts. the earnings declared in household surveys, the whole These estimates can now be made using information of the latter amount is taken. On the basis of this new from household surveys. This is the second methodological vector of self-employed workers’ earnings, it is possible option adopted in this article. The ratio between total to establish a relationship of proportionality between the wage income and total self-employment income is total earnings of self-employed workers and the wage estimated on the basis of microdata from household total (also reported in household surveys). This ratio is surveys, and this ratio is used to correct the total share applied to the sna wage data to reach a final estimate of of wages in gdp. total earnings, which is then compared to gdp.

VI The wage share of total income in Latin America

Information from the countries’ national accounts data by data from the relevant official bodies in cases where can be used for an initial analysis of the evolution of the countries are not incorporated into cepalstat (Argentina, wage share of national income in the region’s countries. Costa Rica and Guatemala). The wage share in Uruguay, As already discussed, this is a rough approximation, as for which sna information only goes up to 2005, was it only includes the earnings of wage workers. Again, updated using the average nominal wage index, the there are large jumps in the time series for the region employment rate and gdp at factor cost. at points where the base year for the national accounts Taking the latest year with information available changes, in view of which the decision was taken to (around 2009), the total wage share ranges from 24% consider the continuous time series available for the in Peru to 56.7% in Costa Rica (see table 1). Analysis countries in the period between 1990 and the latest of the evolution of this share reveals a decline in most year available (which differs by country). The ratio of the countries (8 out of a total of 12), the exceptions between remuneration of labour and gdp at factor cost being the Bolivarian Republic of Venezuela, Chile, Costa is considered.13 cepalstat information is supplemented Rica and Paraguay. Of the countries where there was an improvement in the period considered, Costa Rica stands out as the only one to show a steady upward trend in 13 The measurements yield larger shares at factor cost than at market wages as a share of gdp. In the Bolivarian Republic of prices, since the market price gdp calculation includes product and production taxes minus subsidies. Venezuela, Chile and Paraguay, the wage share of gdp

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 37 shows a decline in the last decade after rising strongly in all three years. The other countries either do not have between 1990 and 2000.14 The countries where the wage information for all three years or change their position share fell between 1990 and the late 2000s followed a in the ranking in one of them (see table A.4). more heterogeneous path. In Argentina and Brazil, a The evolution of the wage share of income depends drop in the 1990s was followed by a partial recovery on differences in real wage and labour productivity in the 2000s. In the Plurinational State of Bolivia, the growth. Figure A.2 of the annex compares changes in real wage share improved in the 1990s and fell in the 2000s. wages and labour productivity in the region’s countries. Colombia, Honduras, Panama and Peru registered In the 1990s, productivity grew by more than real wages declines throughout the period (with quite sharp drops in Argentina, the Bolivarian Republic of Venezuela, over the 1990s in the first three cases). Mexico shows Chile, Guatemala, Panama, Peru, the Plurinational State almost no change between the start and end of the of Bolivia and Uruguay, which explains why the wage period, as an increase in the 1990s was followed by an share fell there in that period, and by less in Colombia, almost symmetrical decline in the 2000s. Developments Costa Rica and Paraguay, which explains why the wage in all years are shown for each country in figure A.1 of share rose there. In Brazil, Mexico and Nicaragua, they the annex. grew at a similar pace, and as a result there were no If the countries are classified into three groups, significant changes in functional income distribution in distinguishing (i) those with a wage share of up to 35%, those countries over the period. In the 2000s, real wages (ii) those with a share of between 35% and 45% and grew faster than labour productivity in Argentina, the (iii) those with a share of over 45%, the country ranking Bolivarian Republic of Venezuela, Brazil, Colombia proves fairly stable, since Mexico and Peru are in the and Costa Rica, which accounts for the rise in the first group, Argentina, Colombia and Paraguay in the wage share over the period. The opposite held true in second and Brazil, Costa Rica and Honduras in the third Chile, Guatemala, Mexico, Panama, Paraguay, Peru, the Plurinational State of Bolivia and Uruguay. A very important change in the region’s social 14 The specific years the table 1 data relate to for each country are presented in table A.3 of the annex. The differences are determined indicators has been a shift away from the worsening by data availability. trend in personal income inequality that was seen to a

TABLE 1 Wages as a share of gdp at factor prices

Wages as a share of gdp Change in the wage share of gdp

Around Around Around 1990-2000 2000-2009 1990-2009 1990 (a) 2000 (b) 2009 (c) (b)-(a) (c)-(b) (c)-(a) Argentina 44.7 40.5 42.9 -4.2 2.4 -1.8 Bolivia (Plurinational State of) 38.2 41.9 34.5 3.8 -7.4 -3.6 Brazil 53.5 47.1 51.4 -6.3 4.3 -2.1 Chile 38.7 46.5 45.4 7.8 -1.1 6.7 Colombia 41.4 36.2 36.1 -5.3 -0.1 -5.3 Costa Rica 48.3 50.6 56.7 2.3 6.1 8.4 Guatemala 36.3 32.8 -3.5 Honduras 54.8 47.5 47.4 -7.3 -0.1 -7.4 Mexico 32.3 34.5 32.2 2.2 -2.3 -0.1 Nicaragua 59.6 56.2 -3.4 Panama 58.6 40.6 35.2 -18.0 -5.4 -23.4 Paraguay* 43.4 59.0 47.2 15.7 -11.9 3.8 Peru 28.7 27.0 24.0 -1.8 -3.0 -4.7 Uruguay 47.4 45.8 -1.6 Venezuela (Bolivarian Republic of) 31.1 35.6 33.5 4.5 -2.1 2.4

Source: prepared by the authors, on the basis of information from cepalstat, the National Institute of Statistics and Censuses (indec) of Argentina, the Central Bank of Costa Rica, the Bank of Guatemala and the Central Bank of Uruguay. * In Paraguay, the 1990 figure is 50% below the average for the indicator in 1991-2009. Including it hugely distorts the evolution of the time series, so the decision was taken to start the analysis for that country in 1991. gdp: gross domestic product.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 38 CEPAL REVIEW 114 • DECEMBER 2014 greater or lesser degree in virtually all the Latin American matched by a greater share for wages in total gdp. countries during the 1990s. After rising for a decade in Income inequality fell between 2002 and 2009 in all the most of them at that time, inequality indicators began Latin American countries except Costa Rica, but only in to decline in many cases in 2002 or 2003, depending on Argentina, Brazil, Costa Rica and Uruguay did the wage the country (see eclac, 2012a and 2012b). The various share of total gdp rise.16 Distributive improvements at studies agree in identifying the role of the labour market the household level have not usually been matched by in this overall decline in household income inequality, a more egalitarian share-out in terms of appropriation since income from the labour market has become less by capital and labour. One theory is that this might be unequal, driving the decline in inequality.15 However, because the income figures on which personal distribution as is revealed by the foregoing analysis and illustrated indicators such as the Gini coefficient are based do not in figure 3, the greater homogeneity of earnings, which fully incorporate property income in practice owing occurred in a context of rising incomes, has not been to issues of data capture. If this is so, it might be said that total wages have been distributed more equitably overall in the last decade, without significant changes 15 One of the main limitations of household surveys is that they (or indeed with greater concentration in asset-owning struggle to capture incomes properly at the top of the distribution. One sectors) in the distribution of the economic surplus option that has gained ground in recent years is the incorporation of generated in the region. other sources of data into the analysis, especially income and wealth data from fiscal records (see Piketty, 2003; Atkinson and Piketty, 2007 and 2010). Studies of this type have been carried out for some countries in the region (see Alvaredo, 2010; Alvaredo and Londoño, 2013; Burdín, Vigorito and Esponda, 2014). Traditional inequality estimates from household survey data have also been corrected using 16 Figure 3 does not include Guatemala, since the latest Gini index national accounts data (see Yamada, Castro and Bacigalupo, 2012). figure available for the country is from 2006.

FIGURE 3 The wage share of gdp and the Gini index, 2002-2009

Wage share of  Gini index

60 0.7 Costa Rica 55 0.65 Brazil 50 Uruguay Paraguay 0.6 45 Honduras Colombia Argentina Chile Brazil 40 0.55 Panama Honduras 2009

2009 Colombia Panama Mexico Chile 35 Bolivia (Plur. State of) Costa Rica Paraguay Bolivia (Plur. State of) Venezuela (Bol. Rep. of) 0.,5 Argentina Mexico Nicaragua 30 Peru 0.45 25 Peru Uruguay Venezuela (Bol. Rep. of) 20 0.4 20 30 40 50 60 0.40 0.45 0.50 0.55 0.60 0.65 0.70

2002 2002

Source: prepared by the authors, on the basis of information from cepalstat, the National Institute of Statistics and Censuses (indec) of Argentina, the Central Bank of Costa Rica and the Central Bank of Uruguay. gdp: gross domestic product.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 39

VII The scale of self-employment in Latin America

One of the distinctive features of labour markets in The income reported by self-employed workers Latin America is the scale of self-employment, which represents a very substantial proportion of total per encompasses own-account workers and employers. capita household income (about 31% in the region). Self-employment accounts for a very large proportion Theoretically, given the nature of the activities involved, of total employment in the region: almost 32% on some of this income is payment for labour and another average in the countries considered, ranging from 22% part is returns to capital. As explained earlier, this income in Argentina to over 49% in Colombia around 2011. is not included in the wage share reported in the national The share of wage employment has increased in the accounts, and this is a major limitation, especially last decade from 59.8% to 63.7% of total employment when it comes to comparing countries with different in the region, and there has also been a small increase degrees of development. Two adjustments to traditional in the wage share of per capita household income (see estimates of the earnings share designed to incorporate table 2). self-employed workers’ incomes will now be presented.

TABLE 2 Wage and self-employment work and incomes

Around 2000 Around 2011

Total employment share Household income share Total employment share Household income share

Self- Self- Wage Wage Wage Wage Self-employed employment Self-employed employment workers income workers income income income Argentina 72.0 26.8 42.4 30.3 76.9 22.4 49.8 25.1 Bolivia (Plurinational State of) 32.1 47.5 41.5 27.7 41.3 40.8 46.8 37.3 Brazil 62.8 31.1 35.4 20.7 68.4 28.7 42.4 19.4 Chile 74.4 24.1 46.4 29.9 77.4 22.3 52.8 26.3 Colombia 49.4 45.6 45.2 27.8 46.0 49.5 44.0 29.7 Costa Rica 71.6 26.5 64.9 19.8 75.9 22.8 62.8 17.3 Guatemala 47.0 38.8 38.2 45.8 51.4 35.0 32.5 42.1 Honduras 49.6 41.2 45.5 29.0 43.7 45.7 47.4 28.5 Mexico 66.1 26.6 46.7 28.4 73.0 22.1 49.7 13.6 Nicaragua 52.3 35.1 50.7 40.8 48.9 39.3 50.4 38.2 Panama 62.7 32.3 60.3 24.5 67.5 28.7 54.5 28.9 Peru 40.5 44.0 39.4 29.8 44.9 42.9 42.5 31.1 Paraguay 44.8 45.2 42.9 37.8 52.9 39.5 47.0 41.7 Uruguay 72.7 25.7 42.1 16.1 71.9 26.8 46.5 16.3 Venezuela (Bolivarian Republic of) 56.4 41.9 45.5 39.1 57.2 41.9 52.3 27.9 Latin America 59.8 33.4 45.4 31.7 63.7 31.6 46.2 30.9

Source: Economic Commission for Latin America and the Caribbean (eclac), on the basis of continuous household survey data.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 40 CEPAL REVIEW 114 • DECEMBER 2014

VIII Reestimation of the labour share including self-employment income

As detailed in subsection V.2, two adjustments were were used 71% of the time. This outcome is reasonable, made to estimate the earnings share of gdp. The first as it presumably reflects the greater importance of returns is the more current in the literature (being similar, for to capital among employers. A new earnings vector was example, to that used in ilo, 2013) and consists in thus estimated for each of the self-employed workers imputing the average wage estimated from sna data to included in the household survey. The ratio between the all self-employed workers and adding this amount to wage total given by the survey and the earnings total sna-reported wages and salaries.17 estimated by this methodology was then calculated. This The second option, detailed in subsection V.2, is ratio was applied to the wage total reported in the sna more rigorous and consists in using information from for each country and year, and total earnings were thus household surveys to estimate the ratio between total estimated (i.e., wage income plus self-employed workers’ wages and total self-employment income, and then earnings) and measured against gdp. Table 3 compares correcting the wage share of gdp on this basis. the three results: the wage share of gdp according to Wage equations were estimated for the totality the sna, the correction arrived at by assuming that self- of wage employees in the economy, with sex, age and employed workers earn roughly the average wage, and age squared, years of education and binary variables the more detailed correction made using the methodology distinguishing branches of activity being taken as the explained above. dependent variables.18 Setting out from the coefficients In all cases, the adjustment made by imputing estimated in these equations, an earnings prediction average wages leads to a large overestimation of total was made for each of the self-employed workers (both earnings in the region’s countries. In the cases of Honduras own-account workers and employers) included in the (2010), Paraguay (2001) and the Plurinational State household surveys. When the predicted earnings of of Bolivia (2000), the result is that earnings amount self-employed workers proved lower than the income to almost 100% of gdp measured at factor cost. When reported by them in household surveys, returns to capital the estimates are carried out by the second method, the were assumed to make up the difference. Accordingly, finding is that a substantial proportion of self-employed the figure yielded by the prediction is taken to be the workers, and own-account workers in particular, declare amount earned by these self-employed workers. If lower earnings in surveys than would be expected from the earnings predicted are greater than the earnings the predictions based on their personal characteristics declared in household surveys, the whole of the latter and branch of activity.19 This income gap relative to amount is taken as the earnings of the self-employed wage workers indicates that simply imputing average workers. The result of following this criterion was that wages produces a substantial bias. On average, this the predictions were used for only 41% of own-account second method of estimation increases the size of wage workers (taking the average across the countries for the earnings by 25 points. years considered), although for employers the predictions

19 A limitation of the adjustment methodology used in this study is its assumption that wage workers and self-employed workers pay 17 Specifically, what is added is the product of the average wage similar amounts of social security contributions and taxes and that multiplied by the number of self-employed workers in the country. The the proportions contributing are likewise similar (given that a ratio estimates for the self-employed worker total come from continuous obtained from the net incomes reported in household surveys is household surveys and cepalstat. applied for most of the countries) to a variable (the wage total) that 18 The results of these estimates are available from the authors on request. includes contributions.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 41

TABLE 3 Total wages and estimated total earnings as shares of gdp

Earnings/gdp Earnings/gdp Wages/gdp Estimate 1 Estimate 2

Argentina 2000 40.5 54.4 45.7 2006 41.5 54.2 48.7 Bolivia (Plurinational State of) 2000 41.9 100.1 56.9 2007 34.5 70.9 47.0 Brazil 2001 47.7 68.2 56.9 2009 51.4 70.6 59.7 Chile 2000 46.5 61.3 54.9 2009 44.5 58.1 52.8 Colombia 2000 36.2 68.3 41.2 2009 44.5 58.1 52.8 Costa Rica 2000 50.6 69.8 55.8 2010 56.9 73.8 65.0 Guatemala 2003 35.5 61.3 46.6 2007 33.5 53.7 44.4 Honduras 2010 47.5 93.1 65.5 Mexico 2000 34.5 48.2 45.6 2008 31.4 41.1 36.2 Panama 2000 40.6 54.4 47.8 2009 35.2 49.2 40.6 Paraguay 2001 58.5 116.7 84.3 2007 47.2 87.7 68.0 Peru 2000 27.0 55.1 35.9 2009 23.3 46.0 31.3 Uruguay 2000 47.4 63.6 57.0 2010 45.8 62.0 54.3 Venezuela (Bolivarian Republic of) 2000 35.6 56.4 46.5 2010 38.4 58.2 44.5

Source: prepared by the authors, on the basis of information from cepalstat, the National Institute of Statistics and Censuses (indec) of Argentina, the Central Bank of Costa Rica, the Central Bank of Uruguay and continuous household surveys. gdp: gross domestic product.

Estimation by the second method yields a substantial The household survey information can be used to increase in the earnings share over the sna figures, analyse the distribution of the total earnings estimated which only include wages, but the results are lower than by the second methodology in accordance with worker those produced by imputing the average wage. This new characteristics, taking the sum of wage incomes and estimation, which is considered the best for the purposes the new own-account worker earnings vector from the of this paper, reflects the volume of earnings, which household survey, by worker education level and sex. range from 31% of gdp in Peru to 65% in Costa Rica. Where education levels are concerned, the distribution Taking the average for all the countries considered, the evinces great stability across countries, with over half of earnings share increases by 10 percentage points over all earnings being generated by workers with complete the figure for wages alone. or incomplete tertiary education (see figure 4). Women Changes in the earnings total are fairly similar across generate an average of 35% of all earnings in the region’s the board, although greater in countries where there is economies, and their share has been rising in most of more self-employment, such as Colombia. the countries analysed (see figure 5).

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 42 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 4 Shares of total earnings by education level, around 2000 and 2009a (Percentages)

70

60 57 5758 5857 55 56 5554 54 54 53 53 52 53 51 51 51 52 50 5050 49 5050 50 48 46 44 42 41 40

32 32 31 30 30 29 29 29 29 30 27 28 27 26 26 25 26 26 26 26 25 24 24 2425 24 23 23 24 24 24 23 2323 2324 23 24 24 22 22 22 21 2121 20 20 20 20 20 1919 18 18 20 16 17 1413 10

0 Primary Primary Primary Primary Primary Primary Primary Primary Primary Primary Primary Primary Primary Primary Primary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary Post-secondary

Argentina Bolivia Brazil Chile Colombia Costa Rica Guatemala Honduras Mexico Panama Paraguay Peru Uruguay Venezuela Latin (Plurinational (Bolivarian America State of) Republic (simple of) average) Around 2000 Around 2009 Source: prepared by the authors, on the basis of continuous household survey data. a The bars show the percentages of total earnings accounted for by workers with complete primary, secondary and post-secondary education, respectively. For the first period, the years other than 2000 are 2001 for Brazil and 2003 for Guatemala. For the second period, the years other than 2009 are 2006 for Argentina, 2007 for the Plurinational State of Bolivia, 2008 for Mexico and 2010 for Costa Rica, Guatemala and Uruguay.

FIGURE 5 Shares of total earnings generated by women, around 2000 and 2009a (Percentages)

50 4444 45 40 39 39 40 40 37 37 38 36 35 34 35.6 34 34 35 33 35 33 31 31 33.0 30 31 3130 2930 30 28 26 25 20 15 10 5 0 Peru Chile Brazil Mexico Panama State of) Uruguay Paraguay Honduras Colombia Argentina Guatemala Costa Rica Republic of) Latin America (simple average) Around 2000 Around 2009 Bolivia (Plurinational Venezuela (Bolivarian

Source: prepared by the authors, on the basis of continuous household survey data. a The bars show the percentages of total earnings accounted for by workers with complete primary, secondary and post-secondary education, respectively. For the first period, the years other than 2000 are 2001 for Brazil and 2003 for Guatemala. For the second period, the years other than 2009 are 2006 for Argentina, 2007 for the Plurinational State of Bolivia, 2008 for Mexico and 2010 for Costa Rica, Guatemala and Uruguay.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 43

IX Concluding remarks

Analysis of income distribution currently centres on agendas is a challenge in the region, partly because of inequality between households and individuals, with a the limitations of the information available. Doing so, particular focus on the microeconomic fundamentals of however, can reveal new facets of the distribution situation its evolution. This article has argued that it is important to in the region. The decline in income inequality between retain the functional perspective and seek to comprehend households that the region has been experiencing for a the reality by considering and relating the two approaches. decade has not been matched by improvements in the Integrating functional income distribution into research share of the proceeds of growth appropriated by workers.

ANNEX

TABLE A.1 Information available at eclac

Country Period rl os cfk gdpfc (t-s)xm gdpmp Bolivia (Plurinational State of) 1988/2008 Brazil 1970-1975-1980-1985a 1990/2009 Chile 1960/1985 1985/1996 1996/2006 2003/2009 2008/2010 Colombia 1970/1994 1994/2000 2000/2010 Costa Rica 1970/1991 Ecuador 1970/1989 Honduras 1950/1995 1996/2000 2000/2011 Mexico 1970/1980 1980/1988 1988/2003 2003/2011 Nicaragua 1994/2011 Panama 1960/1970 1970/1980 1980/1996 1996/2011 Paraguay 1970/1991 1991/2007 Peru 1991/2010 Uruguay 1971/1983 1983/1988 Venezuela (Bolivarian Republic of) 1970/1984 1984/1997 1997/2011b

Source: prepared by the authors, on the basis of cepalstat information. a Hyphens do not denote a range of years but are only separators for the years for which information is available. b There are no data for 2010. Note: rl: remuneration of labour; os: operating surplus; cfk: consumption of fixed capital;gdp fc: gross domestic product at factor cost; (t-s)xm: taxes on production and imports less subsidies; gdpmp: gross domestic product at market prices.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 44 CEPAL REVIEW 114 • DECEMBER 2014

TABLE A.2 Information available at unsd

Country Period rl mi os cfk (t-s)xm gvabp Argentina 1993/2007 Bolivia (Plurinational State of) 1970/2011 Brazil 1992/2003 1995/2009 Chile 1974/1985 1985/1998 1996/2009 2008/2010 Colombia 1970/1995 1992/2005 2000/2010 Costa Rica 1970/1993 1991/2010 Ecuador 1970/1991 Guatemala 2001/2010 Honduras 1992/2006 2000/2010 Mexico 1988/2004 1993/2004 2003/2010 Nicaragua 1994/2007 Panama 1989/2000 1996/2010 Paraguay 1994/2010 Peru 1970/1998 1991/2010 Dominican Republic 1991/2005 Uruguay 1997/2005 Venezuela (Bolivarian Republic of) 1970/1984 1984/2002 1997/2010

Source: prepared by the authors, on the basis of data from the United Nations Statistics Division (unsd). Note: rl: remuneration of labour; mi: mixed income; os: operating surplus; cfk: consumption of fixed capital: t-s( )xm: taxes on production and imports less subsidies; gvabp: gross value added at basic prices.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 45

TABLE A.3 Actual years for country data in table 1

Around 1990 Around 2000 Around 2009 Argentina 1993 2000 2007 Bolivia (Plurinational State of) 1990 2000 2007 Brazil 1991 2000 2009 Chile 1990 2000 2009 Colombia 1990 2000 2009 Costa Rica 1991 2000 2009 Guatemala … 2001 2009 Honduras 1990 2000 2009 Mexico 1990 2000 2009 Nicaragua 1994 2000 n/a Panama 1990 2000 2009 Paraguaya 1991 2000 2009 Peru 1990 2000 2009 Uruguayb n/a 2000 2009 Venezuela (Bolivarian Republic of) 1990 2000 2009

Source: prepared by the authors. a In Paraguay, the 1990 figure is 50% below the average for the indicator in 1991-2009. Including it hugely distorts the evolution of the time series, so the decision was taken to start the analysis for that country in 1991. b The information from the Central Bank of Uruguay runs up to 2005, so the ratio was updated using the evolution of the average nominal wage index, the employment rate and gross domestic product (gdp) at factor cost.

TABLE A.4 Country ranking by wage share of gdp Wages/gdp Around 1990 Around 2000 Around 2009

Up to 35% Peru, Mexico, Bolivarian Mexico, Peru Plurinational State of Bolivia, Republic of Venezuela Guatemala, Mexico, Peru, Bolivarian Republic of Venezuela 35%-45% Argentina, Plurinational State Guatemala, Argentina, Bolivia, Argentina, Colombia, Panama of Bolivia, Chile, Colombia, Colombia, Panama, Bolivarian Paraguay Republic of Venezuela Over 45% Brazil, Costa Rica, Honduras, Brazil, Chile, Costa Rica, Brazil, Chile, Costa Rica, Nicaragua, Panama Honduras, Nicaragua, Honduras, Paraguay, Uruguay Paraguay, Uruguay

Source: prepared by the authors. gdp: gross domestic product.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 46 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE A.1 Latin America (15 countries): wages as a share of gdp

Argentina Bolivia (Plurinational State of) Brazil 60 60 60 55 55 55 50 50 50 45 45 45 40 Index 1993 = 100 40 40 35 35 35 Index 1991 = 100 30 Index 1990 = 100 30 30 25 25 25 20 20 20 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Chile Colombia Costa Rica 60 60 60 55 55 55 50 50 50 45 45 45 40 40 40 35 35 35 Index 1990 = 100 Index 1991 = 100

30 Index 1990 = 100 30 30 25 25 25 20 20 20 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Guatemala Honduras Mexico 60 60 60 55 55 55 50 50 50 45 45 45 40 40 40 35 35 35 Index 2001 = 100 30 Index 1990 = 100 30 Index 1990 = 100 30 25 25 25 20 20 20 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Nicaragua Panama Paraguay 60 60 60 55 55 55 50 50 50 45 45 45 40 40 40 35 35 35 Index 1994 = 100 Index 1991 = 100

30 Index 1990 = 100 30 30 25 25 25 20 20 20 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Peru Uruguay Venezuela (Bolivarian Republic of) 60 60 55 60 55 50 55 50 45 50 45 40 45 40 35 40 35 Index 1990 = 100 30 35 Index 1990 = 100 Index 1997 = 100 30 25 30 25 20 25 20 20 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Source: prepared by the authors, on the basis of information from cepalstat, the National Institute of Statistics and Censuses (indec) of Argentina, the Central Bank of Costa Rica, the Bank of Guatemala and the Central Bank of Uruguay. gdp: gross domestic product.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 47

FIGURE A.2 Latin America (14 countries): real wages and labour productivity, 1990-2010

Argentina Bolivia (Plurinational State of) Brazil 150 150 150

130 130 130

110 110 110

90 90 90 Index 1990 =100 Index 1991 = 100 Index 1993 = 100 70 70 70

50 50 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Wage index /worker index Wage index /worker index Wage index /worker index

Chile Colombia Costa Rica

190 150 150 170 130 130 150 110 110 130 90 90 110 Index 1990 = 100 Index 1991 = 100 Index 1990 = 100 70 90 70

70 50 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Wage index /worker index Wage index /worker index Wage index /worker index

Guatemala Mexico Nicaragua

150 150 150

130 130 130

110 110 110

90 90 90 Index 1990 = 100 Index 2001 = 100 70 70 Index 1994 = 100 70

50 50 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Wage index /worker index Wage index /worker index Wage index /worker index

Panama Paraguay Peru 250 150 190 130 170 200 150 110 130 150 90 110 100 90 Index 1990 = 100 Index 1990 = 100 Index 1991 = 100 70 70 50 50 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Wage index /worker index Wage index /worker index Wage index /worker index

Uruguay Venezuela (Bolivarian Republic of)

130 130 120 120 110 110 100 100 90 90 80 80 70 70 Index 1997 = 100 Index 1990 = 100 60 60 50 50 40 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Wage index /worker index Wage index /worker index

Source: prepared by the authors on the basis of information from cepalstat, the National Institute of Statistics and Censuses (indec) of Argentina, the Central Bank of Costa Rica, the Bank of Guatemala and the Central Bank of Uruguay. gdp: gross domestic product.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA 48 CEPAL REVIEW 114 • DECEMBER 2014

Bibliography Abeles, M. and F. Toledo (2011), “Distribución del ingreso y análisis Ellis, L. and K. Smith (2010), “The global upward trend in the macroeconómico: Un repaso de la literatura y de los desafíos profit share”,Applied Economics Quarterly, vol. 56, No. 3, de política económica”, Distribución del ingreso. Enfoques y Duncker & Humblot. políticas públicas desde el Sur, M. Novick and S. Villafañe European Commission (2007), The Labour Income Share in (eds.), Buenos Aires, United Nations Development Programme the European Union, Brussels, Directorate General for (undp)/Ministry of Labour, Employment and Social Security Employment, Social Affairs and Equal Opportunities. of Argentina. Feldstein, M.S. (2008), “Did wages reflect growth in productivity?”, Alvaredo, F. (2010), “The rich in Argentina over the twentieth nber Working Paper, No. 13953, Cambridge, Massachusetts, century”, Top Incomes: A Global Perspective, A. Atkinson and National Bureau of Economic Research. T. Piketty (eds.), Oxford University Press. Fichtenbaum, R. (2009), “The impact of unions on labor’s share of Alvaredo, F. and J. Londoño (2013), “High incomes and personal income: a time-series analysis”, Review of Political Economy, taxation in a developing economy: Colombia 1993-2010”, vol. 21, No. 4, Taylor & Francis. unpublished. Goldfarb, R.S. and T.C. Leonard (2005), “Inequality of what Amarante, V. and A. Vigorito (2011), “Los futuros posibles de la among whom?: Rival conceptions of distribution in the 20th desigualdad de ingresos”, La aventura uruguaya. ¿Naides century”, Research in the History of Economic Thought and más que naides?, R. Arocena and G. Caetano (eds.), Methodology, vol. 23, Part 2, Emerald. Montevideo, Debate. Gollin, D. (2002), “Getting income shares right”, Journal of Antras, P. (2004), “Is the U.S. aggregate production function Political Economy, vol. 110, No. 2, Chicago, University of Cobb- Douglas? New estimates of the elasticity of Chicago Press. substitution”, Contributions in Macroeconomics, vol. 4, No. 1, Graña, J.M. (2007), Distribución funcional del ingreso en la Argentina: Harvard University. 1935-2005, Buenos Aires, University of Buenos Aires. Atkinson, A. (2009), “Factor shares: the principal problem of political Harrison, A. (2002), “The World Bank: structure and policies: economy?”, Oxford Review of Economic Policy, vol. 25, Christopher L. Gilbert and David Vines (eds.), Cambridge No. 1, Oxford University Press. University Press”, Journal of International Economics, Atkinson, A. and T. Piketty (2010), Top Incomes: A Global vol. 57, No. 1, Amsterdam, Elsevier. Perspective, Oxford University Press. Hernández Laos, E. (1998), “Políticas de estabilización y ajuste (2007), Top Incomes over the Twentieth Century: A y distribución funcional del ingreso en México”, Revista Contrast between Continental European and English-Speaking Comercio Exterior, Mexico City, Banco Nacional de Countries, Oxford University Press. Comercio Exterior. Bentolila, S. and G. Saint-Paul (2003), “Explaining movements in Hogrefe, J. and M. Kappler (2012), “The labour share of income: the labor share”, The B.E. Journal of Macroeconomics, vol. 3, heterogeneous causes for parallel movements?”, The Journal No. 1, De Gruyter. of Economic Inequality, vol. 11, No. 3, Springer. Bernanke, B.S. and R.S. Gürkaynak (2002), “Is growth ilo (International Labour Organization) (2013), Global Wage Report exogenous? Taking Mankiw, Romer, and Weil seriously”, 2012/2013. Wages and Equitable Growth, Geneva. nber Macroeconomics Annual 2001, vol. 16, Cambridge, (2011), Global Wage Report 2010/2011. Wage Policies in Massachusetts, National Bureau of Economic Research. Times of Crisis, Santiago, Chile. Bhaduri, A. (1983), The Economic Structure of Backward Agriculture, (2008), World of Work Report 2008. Income Inequalities Academic Press. in the Age of Financial Globalization, Geneva. Bhaduri, A. and S. Marglin (1990), “Unemployment and the real imf (International Monetary Fund) (2007), Annual Report 2007, wage: the economic basis for contesting political ideologies”, Washington, D.C. Cambridge Journal of Economics, vol. 14, No. 4, Oxford Jayadev, A. (2007), “Capital account openness and the labour University Press. share of income”, Cambridge Journal of Economics, vol. 31, Burdín, G., A. Vigorito and F. Esponda (2014), “Desigualdad No. 3, Oxford University Press. y sectores de altos ingresos en Uruguay: Un análisis Kaldor, N. (1961), Capital Accumulation and Economic Growth, en base a registros tributarios y encuestas de hogares MacMillan. para el período 2009-2011”, Documento de Trabajo, No. Krueger, A. (1999), “Measuring labor’s share”, nber Working Paper, 06/04, Montevideo, Institute of Economics, University of No. 7006, Cambridge, Massachusetts, National Bureau of the Republic. Economic Research. Campodónico, H. (2008), “Renta petrolera y minera en países Lewis, W.A. (1954), “Economic development with unlimited supplies seleccionados de América Latina”, Project Document, of labour”, The Manchester School, vol. 22, No. 2, Wiley. No. 188 (LC/W.188), Santiago, Chile, Economic Commission Lindenboim, J. (2008), “Distribución funcional del ingreso, un tema for Latin America and the Caribbean (eclac). olvidado que reclama atención”, Problemas del Desarrollo. Daudey, E. and C. García-Peñalosa (2007), “The personal and the Revista Latinoamericana de Economía, vol. 39, No. 153, factor distributions of income in a cross-section of countries”, Mexico City. The Journal of Development Studies, vol. 43, No. 5, Taylor Lindenboim, J., D. Kennedy and J.M. Graña (2010), “El debate sobre & Francis. la distribución funcional del ingreso”, Desarrollo Económico, eclac (Economic Commission for Latin America and the Caribbean) vol. 49, No. 196, Buenos Aires, Institute of Economics and (2013), Natural Resources within the Union of South American Social Development. Nations. Status and Trends for a Regional Development Agenda Ministry of Planning and Cooperation (2000), Estudio sobre la (LC/L.3627), Santiago, Chile. distribucion del ingreso: Estructura funcional en 1987-96 y (2012a), Structural Change for Equality: An Integrated proyecciones, Santiago, Chile. Approach to Development (LC/G.2524(SES.34/3)), Piketty, T. (2003), “Income inequality in France, 1901-1998”, Journal Santiago, Chile. of Political Economy, vol. 111, No. 5, Chicago, University of (2012b), Social Panorama of Latin America 2011 Chicago Press. (LC/G.2514-P), Santiago, Chile. United Nations publication, Ricardo, David (1973), Principios de economía política y tributación, Sales No. E.12.II.G.6. Madrid, Editorial Ayuso.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA CEPAL REVIEW 114 • DECEMBER 2014 49

Rodríguez, O. (2006), El estructuralismo latinoamericano, Santiago, analysis of the determinants of functional income distribution”, Chile, Economic Commission for Latin America and the Conditions of Work and Employment Series, No. 35, Geneva, Caribbean (eclac). International Labour Organization (ilo). Rodríguez, F.R. and D. Ortega (2006), “Are capital shares higher in Taylor, L. (1991), Income Distribution, Inflation, and Growth: poor countries? Evidence from industrial surveys”, Wesleyan Lectures on Structuralist Macroeconomic Theory, Cambridge, Economics Working Papers, No. 2006-023, Middletown, Massachusetts, The mit Press. Wesleyan University. undp (United Nations Development Programme) (2010), Human Serrano, F. and C. Medeiros (2001), Economic Development and Development Report 2010. The Real Wealth of Nations: the Revival of the Classical Surplus Approach, Cape Town. Pathways to Human Development, New York. Serres, A., S. Scarpetta and C. Maisonneuve (2001), “Falling wage Yamada, G., J. Castro and J. Bacigalupo (2012), “Desigualdad shares in Europe and the United States: how important is monetaria en un contexto de rápido crecimiento económico. El aggregation bias?”, Empirica, vol. 28, No. 4, Springer. caso reciente de Perú”, Documento de Discusión, No. 12/01, Solow, R.M. (1958), “A skeptical note on the constancy of relative University of the Pacific Research Centre. shares”, The American Economic Review, vol. 98, No. 4, Young, A. (1995), “The tyranny of numbers: confronting the statistical Nashville, Tennessee, American Economic Association. realities of the East Asian growth experience”, The Quarterly Stockhammer, E. (2013), “Why have wage shares fallen? A panel Journal of Economics, vol. 110, No. 3, Oxford University Press.

THE EARNINGS SHARE OF TOTAL INCOME IN LATIN AMERICA, 1990-2010 • MARTÍN ABELES, VERÓNICA AMARANTE AND DANIEL VEGA

Latin America: Total factor productivity and its components

Jair Andrade Araujo, Débora Gaspar Feitosa and Almir Bittencourt da Silva

ABSTRACT This article applies the stochastic-frontier model to examine total factor productivity (tfp) and its components in Latin America between 1960 and 2010. The likelihood-ratio test shows that, for a selection of Latin American countries over the 50 years analysed, the macroeconomic variables of technical inefficiency included in the model generally have a significant effect; and they allow for a better understanding of technical inefficiency throughout the region. The key variables explaining technical inefficiency in the selected countries are public expenditure and the inflation rate; and there is also an inverse relation between technical inefficiency and the extent to which local prices diverge from purchasing power parity.

KEYWORDS Productivity, measurement, mathematical analysis, econometric models, Latin America

JEL CLASSIFICATION O47, O54, O57

AUTHORS Jair Andrade Araujo has a Ph.D. in Economics and is a Professor on the Masters Course in Rural Economics (maer) of the Federal University of Ceará (ufc), Brazil. [email protected] Débora Gaspar Feitosa has a Ph.D. in Economics and is a Professor on the Economic Sciences and Finance Courses of the Federal University of Ceará (ufc), Brazil. [email protected] Almir Bittencourt da Silva has a Ph.D. in Economics and is a Professor on the Postgraduate Course in Economics (caen) of the Federal University of Ceará (ufc), Brazil. [email protected] 52 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

Using the concept of total factor productivity (tfp), of this approach is that allows tfp to be broken down quantified using the Cobb-Douglas production function, into components that characterize the general production Solow (1957) introduced the measurement of the process. The procedure used makes it possible to identify contribution made by technical progress to per capita the components of technical efficiency, which reflect the output growth. This author estimated the production way an economy moves towards the production frontier, function of the United States economy from 1909 to 1949, as distinct from the technical progress component, which and established the existence of a residual, measured as represents a shifting of the frontier itself. the difference between the growth rates of real output An advantage of the procedure used by Bauer (1990) and the weighted growth rates of the individual factors and Kumbhakar (2000) is that by permitting a flexible of production, capital and labour. The importance of specification of the production frontier, such as translog, technical progress, which was discovered when attempting tfp can be broken down into the components of technical to break down the growth rate of real output using the efficiency, allocative efficiency, scale effect and technical growth rates of the factors of production, is known as progress. This procedure is superior to decomposition the Solow residual. using the tfp Malmquist index (based on a production The notion of technical progress then came to be frontier that is restricted by the imposition of constant used as an abbreviated expression for any shift of the returns to scale), which is used in many other studies. production function. Nonetheless, based on empirical In this case, according to Färe and others (1992), tfp is studies founded on growth accounting and inspired in the divided into just two elements: the variation in technical neoclassical model, some authors started to claim that efficiency and technological change. This line of research several causes could be closely related to the measurement also includes studies by Kumbhakar and Lovell (2003); of the residual. Drawing on Solow’s work, a number of Sauer, Frohberg and Hockmann (2006), and Henningsen empirical studies were performed —such as those and Henning (2009). of Griliches (1996), which uses various methodologies This article uses the stochastic-frontier model and samples— aimed at analysing the components of the to analyse the contribution made by tfp to economic aforementioned residual, with a view to quantifying as growth in a sample of Latin American countries; for precisely as possible the real contribution that technical which purpose it examines the components of technical progress makes to output growth. efficiency, allocative efficiency, scale effect, and technical Orea (2002) studies panel data based on information progress in the variation of tfp in those countries. It from Spanish banks and proposes a parametric breakdown thus contributes to the empirical literature for a better of the Malmquist index. The results show that tfp growth understanding of the real factors that underlie the can be attributed mainly to technical progress. Färe, economic performance of the sample countries over Grosskopf and Roos (1998) also study productivity and a 50-year period. A further aim is to understand the the Malmquist index. The empirical studies of Färe and influence of the vector of macroeconomic variables on others (1994); Johnson and Kuosmanen (2012); Lee the technical efficiency of the countries in the sample, and others (2013), and Wang and others (2014), among through the technical inefficiency model, following various authors, also show that it is possible to study the Battese and Coelli (1995). productivity of economic agents using non-parametric or The article is divided into six sections including this semi-parametric methods. For example, data envelopment introduction. Section II briefly explains the stochastic- analysis (dea) is a non-parametric methodology that can frontier model and the tfp decomposition procedure; and be used to evaluate the technical efficiency of productive section III presents the databases, sample of countries and units and estimate the Malmquist index. the econometric model used. Section IV demonstrates The present article reports an application of the the calculation of the tfp decomposition, using the tfp decomposition procedure suggested by Bauer Bauer (1990) and Kumbhakar (2000) procedure; and (1990) and Kumbhakar (2000) to a sample of Latin section V presents the results of the estimation and American countries for the period 1960-2010, based on breakdown. The sixth and last section puts forward a the stochastic production-frontier model. The advantage number of final thoughts.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 53

II Stochastic frontier and tfp decomposition

2 2 This study uses stochastic production-frontier analysis, variance of σ , (v ~ iid N (0, vv )); whereas the second which is one of the methods adopted in the technical- represents technical inefficiency, in other words the part inefficiency literature, to identify one of the components that constitutes a retreat from the production frontier, of tfp, namely technical efficiency. which can be inferred from the negative sign and the The approach uses econometric (parametric) constraint u ≥ 0. These are non-negative random variables techniques, whose production-frontier models are used with a zero-truncated normal distribution, independently to study technical inefficiency, and it is recognized that distributed (not identically) with mean μit and constant 2 2 output can be affected by random disturbances that variance vu ; in other words (u ~ NT (μ,vu )). As the are outside producer control. Unlike non-parametric error components are mutually independent and xit is approaches, which assume deterministic frontiers, assumed exogenous, the model can be estimated using stochastic-frontier analysis allows for deviations from the maximum-likelihood technique. the frontier, for which the error can be broken down into Unlike the model used by Pires and Garcia (2004), changes in technical efficiency and random disturbances. this one has the advantage of allowing inefficiencies and In the deterministic-frontier models, deviations from input elasticities to vary through time, which makes it the production frontier are attributed to the producer’s easier to identify changes in the production structure. technical inefficiency; but those models ignore the fact The effects of technical inefficiency (eit) are expressed that production can be affected by random disturbances with the following characteristics: outside producer control, such as strikes or environmental conditions, among others. eit = zitδ + wit Stochastic-frontier analysis originated in articles by Aigner, Lovell and Schmidt (1977) and Meeusen where: zit is a vector of explanatory variables of the and Broeck (1977), which were followed by the work technical inefficiency of the i-th productive unit (country) of Battese and Corra (1977). Those original studies measured in time t; δ is a vector of parameters associated present the structurally composed error term in the with the variables zit; and wit is a normally distributed 2 context of the production function. Since then, various random variable with zero mean and variance vw . It is authors have collaborated, including Battese and assumed that eit has a zero-truncated normal distribution, Coelli (1995), who modelled technical inefficiency as so its mean is wit = zitδt . time-variant, formalizing the technical inefficiency of Under this formulation, a functional form is defined, the stochastic production function for panel data. The as presented below; and this is used to obtain tfp, which present article adopts the model proposed by Battese is then broken down into its components. and Coelli (1995) and Coelli, Rao and Battese (1998). The decomposition of tfp through the well- Accordingly, the stochastic production-frontier model known Malmquist index, which separates the total can be described through equation (1), where yit is the productivity index into technological variation and vector of quantities produced by the various countries efficiency variation, has been widely used by a number in period t; xit is the vector of factors of production used of authors. For example, Laborda, Sotelsek and Guasch in period t, and β is the vector of parameters defining (2011) study productivity growth in 16 Latin American the production technology. countries between 1996 and 2006, using a stochastic- frontier approach, and they perform a decomposition of the Malmquist index. Nonetheless, in a more wide yfit=−tx,, it b ..expvuit exp it ,u $ 0 __ii_ i (1) ranging study, Bauer (1990) and Kumbhakar (2000) iNtT==11,..., ,,..., propose a type of decomposition which, in addition to the components listed above, captures production- The terms vit and uit are vectors that represent scale effects and changes in the inefficiency of factor different error components. The first relates to the allocation. Section III contains a detailed analysis of random part, with a truncated normal distribution, how to incorporate these components into analyses of independent and identically distributed, with a constant tfp variations.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA 54 CEPAL REVIEW 114 • DECEMBER 2014

III Methodology

1. Description of the sample and the data used hypothesis about the distribution of technical inefficiency (Battese and Coelli (1995)). The analysis considers the following 19 Latin American Firstly, one model was tested using the Cobb- countries: Argentina, the Bolivarian Republic of Douglas functional form and another using the translog Venezuela, Brazil, Chile, Colombia, Costa Rica, the form. The adequacy test identified the latter functional Dominican Republic, Ecuador, El Salvador, Guatemala, form as superior in terms of data consistency. Honduras, Jamaica, Mexico, Nicaragua, Paraguay, Peru, Accordingly, the translog production-frontier the Plurinational State of Bolivia, Trinidad and Tobago, function for the 19 selected Latin American countries and Uruguay. The data relate to the period 1960-2010, was specified as follows: and were obtained from the following sources: Penn World Table 7.1 (pwt 7.1) and the World Development 1 LnYtaa aatL2 nKa tLnK Indicators published by the World Bank. The availability it =+i 2 ++3 2 45it + it of information from those databases was decisive in 1 2 1 2 (2) choosing 2010 as the upper bound of the sample. ++aa67LnLtit LnLLit ++aa89 LnKLit n it 2 __ii2 The variables gross domestic product (pib), ++a10 LnKLitnL itvu it− it labour (l), government consumption expenditure (g) and deviation of local prices from purchasing power parity where: (dppa) were taken from Penn World Table 7.1. The series Y = gdp of country i in period t. for physical capital (k) of the individual countries was it K = physical capital stock of country i in period t. constructed from estimates based on gross investment, it L = labour in country i in period t. using the perpetual-inventory technique. it a = fixed effects, with the aim of capturing unobserved The inflation rate data came from the World i heterogeneities in the sample of countries. Development Indicators, although the difficulties in t = linear trend. obtaining data for certain countries meant that other 1 t2 = quadratic trend. sources were also used. In the case of Brazil, the general 2 price index-domestic supply (igp-di) published by the vit = random disturbances of the production function, Getulio Vargas Foundation was used. which are assumed to be normally distributed The sample consists of annual data on capital, labour, with zero mean and constant variance. gdp, government expenditure, purchasing power parity uit = technical inefficiency of production, modelled and inflation from the 19 selected countries, totalling as follows: 950 observations forming a balanced panel. uzit=+d~ it it (3) 2. Econometric model where: Total factor productivity was calculated using the zit = (z1t, z2t, z3t, z4t) represents a vector of variables stochastic production-frontier model proposed by that explain technical inefficiency, andδ is a Aigner, Leobel and Schmidt (1977) and Meeusen and parameter of associated with zit. 2 Broeck (1977), which was subsequently improved by ωit = assumed normally distributed N(0,v~ ). Pitt and Lee (1981) and Schmidt and Sickles (1984). Under the foregoing hypothesis, it is also assumed This makes it possible to model the panel data with the that uit is independently distributed in a truncated normal productive technical-inefficiency component, according distribution with mean wit = δzit, and constant variance 2 to the foundations used by Battese and Coelli (1995), of v~ . which suggest that technical inefficiency is modelled The choice of the variables used to model the by a vector of variables. technical-inefficiency term is based on a number of This being the case, a functional form of the empirical studies that use this procedure to estimate production frontier is modelled, in conjunction with a parametric frontiers involving aggregate data.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 55

The inefficiency variables considered are as follows: Douglas is the appropriate form for presenting the data z1t = trend effect. is tested, given the translog specifications. This can be z2t = government consumption expenditure relative to tested using the likelihood-ratio test. The table published the gdp of each country. Some empirical studies in Kodde and Palm (1986) is used to compare the critical have been conducted to quantify the effect of values of the results, given the degrees of freedom. The current expenditure outgoings on inefficiency. For test proceeds as follows: example, Bittencourt and Marinho (2007) analyse After obtaining the two models and their respective tfp in Latin American countries and discuss the maximum-likelihood ratios (ll), the generalized- effects of macroeconomic variables in explaining likelihood statistic (lr) of the estimated production the technical inefficiency component through the functions is considered. Then the hypothesis test is applied: stochastic frontier. They find that current government H0: LL Cobb-Douglas. expenditure contributed to increasing technical H1: LL translog. inefficiency in the countries of the sample between and, consequently, the generalized-likelihood ratio, 1961 and 1990. Thus, an increase in government LR = - 2 [ln LL H0 – Ln LL H1] expenditure would be expected to make production LR > T KP (Kodde and Palm, 1986 table) H0 is more technically inefficient. rejected. z3t = corresponds to the logarithm of 1 + the inflation To find an ideal model to represent the data, rate, p, in other words ln (1 + p). This expression further functional form tests were conducted in is used because it captures the non-linear effects addition to that described above between Cobb- of inflation on technical inefficiency. According to Douglas and translog. These tests only changed some De Gregorio (1992), some countries experienced of the inefficiency variables, but owing to a lack of periods of deflation and hyperinflation, but the convergence between certain models, it was impossible to influence of those extreme situations on the make comparisons. inefficiency term is attenuated by using the expression indicated above. Inflation is expected to increase (b) Absence of technical progress the technical inefficiency of production. This test considers whether or not the coefficients z4t = corresponds to the deviation of the local price level of the time-related variables in the translog function from purchasing power parity, using the United are equal to zero. In other words, it tests the hypothesis States as the benchmark country. This variable serves that a2, a3, a5, a7 in equation (2) are equal to above all to control for the technical-inefficiency zero. Thus: effects of trade policies that involve devaluation of H0: a2, a3, a5, a7 = 0. the real exchange rate. H1: complete translog. The parameters of equations (2) and (3) are estimated Using the generalized-likelihood ratio, using the maximum-likelihood method, which makes LR = - 2 [ln LL H0 – Ln LL H1] it possible to calculate the magnitude of technical LR > T KP (Kodde and Palm, 1986 table) H0 is efficiencies for each country in the sample. rejected.

3. Tests conducted (c) Effect of technical inefficiency on the production function (a) Functional form This case tests for the nonexistence of technical Firstly the Cobb-Douglas and then the translog inefficiency; in other words, whether in fact the inefficiency production functions were estimated, for the purpose variables depend on the model. For that purpose, the of using the adequacy test to choose which functional log-likelihood value (ll) is taken of the model estimated form to use in the study. Although the Cobb-Douglas without these variables, and the generalized-likelihood functional form is commonly used in frontier-estimation test is performed again, comparing it with the critical models, it is a simple model with few properties, including value of Kodde and Palm (1986). The degrees of freedom constant elasticity and returns to scale (Coelli, Rao and correspond to the inefficiency variables. Battese, 1998). Thus: Thus, in line with several studies, the functional H0: nonexistence of technical inefficiency. form test is used to estimate both the Cobb-Douglas and H1: alternative hypothesis: the technical inefficiency the translog forms; and the null hypothesis that Cobb- must be taken into account in the model.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA 56 CEPAL REVIEW 114 • DECEMBER 2014

(d) Absence of fixed effects with reference to the critical value of Kodde and This test is used to evaluate the model without Palm (1986). the presence of fixed effects captured by its dummy In this particular study, the estimation without variables. The model is once again estimated without fixed effects did not converge after a large number of taking account of the presence of those dummy iterations, so the model could not be estimated and was variables, and the generalized-likelihood test is applied, rejected for comparison purposes.

IV Decomposition of tfp

1. Composition of the data yc KccL g s s (4) PTFK=−y K − L L In order to break the tpf down into its components, the country data were used to develop the initial econometric in the deterministic part, it can be seen that: model, along with the data calculated from that model. The 19 countries of the sample were maintained for the econometric model, and the period 1960-2010 yc 2 ft,,KL,ln b KccL 2u _ i ff (5) was maintained for the analysis, with data on capital (k), y =+2t KLK +−L 2t labour (l) and gdp (y) mainly being used. The factor shares s and s were obtained from calculations based K L where: on Penn World Table 7.1 data. S = the capital share of income; S = the labour The elasticities and were calculated on the basis K L ek eL share of income; = the elasticity of capital; and of the respective derivatives of the translog production eK = the elasticity of labour. function used in relation to the corresponding factors eL Returns to scale (rts) are defined as the sum of of production. elasticities, such that:

2. Decomposition procedure RTS = eK + eL where, Bauer (1990) and Kumbhakar (2000) proposed a g = growth rate of K productivity decomposition that goes beyond changes in K g = growth rate of L productivity to capture the effects of technical innovation. L This approach also takes account of production-scale fK fL effects. That decomposition is performed by firstly Setting, m and m , and substituting K = RTS L = RTS estimating the model of equations (2) and (3), after in the index, a number of algebraic operations are then which it is possible to “compose” the rate of change of performed: tpf based on the results. According to that model, which was used by Pires gPPTFK=−Tuc +−RTSg1 ..m K + mLL.g and Garcia (2004), on the basis of the formulation _ i 7 A (6) proposed by Battese and Coelli (1993), it is possible +−mmKKsg..KL+−sgLL 9__iiC to study the effects of each tpf component. The main advantage of this is that it admits the possibility of Equation (6) describes the rate of variation of tfp, variable returns to scale. gPTF, which can be broken down into four elements: In this way, the components of productivity technical progress, variation of technical efficiency, can be identified after some algebraic manipulations variations in the scale of production, and variations in on the expression that represents the deterministic the efficiency of resource allocation. part of the production frontier. Pires and Garcia Technical progress (pt) is represented by the (2004) present an index for the tpf growth rate derivative of the production function with respect expressed as: to time:

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 57

2 ft,,KL,ln b one separately. For example, if technology does not PT _ i change (if PT = 0 in the item defined above), this will = 2t not contribute to productivity gains. Similarly, technical The change in technical efficiency is denoted by the inefficiency, which changes through time, will have technical inefficiency coefficient with a negative sign −uc. repercussions on the rate of variation; otherwise, if It should be noted that the change in production −=uc 0= 0, it will not affect that rate. scale is given by the expression that contains the With constant returns to scale (RTS = 1), the third returns to scale and growth rates of capital and component of the formula for the variation in productivity labour, in other words the third term of equation (6): is zero. Nonetheless, if RTS ≠ 1 productivity can partially (RTS -1).[λK.gk + λL.gL]. explain returns to production scale. Changes relating to allocative efficiency are By setting λK + λL = 1, one can discern a symmetry represented by the last term in equation (6), which relates in the distances of the shares of k and l with respect proportionate returns to scale, the shares of capital and to λ, where the capital and labour shares are symmetric labour, and the growth rates, and is thus measured by: and, consequently, have opposite signs. According to a [(λK - sk).gK + (λL - sL ).gL]. reallocation factor, this means that the intensity of one This methodology, with which productivity is factor will reduce the intensity of the other; in other broken down into the four components mentioned, words, that capital intensity will reduce the amount of makes it possible to evaluate the repercussion of each labour, and vice versa.

V Estimation and results

1. Analysis of the estimation of the are statistically significant at the 1% level, and have the production frontier expected signs. For example, the estimated coefficient of the trend Table 1 shows the model that corresponds to the variable (z1t) in the technical inefficiency model has a estimation of the production frontier in the translog positive sign and is statistically significant at 1%, which functional form, which was the model that best fit the could indicate a tendency for inefficiency to increase in data after the appropriate tests described above. All of the period studied. the estimated parameters are statistically significant at The government current expenditure variable (z2t) the 5% level, except for the a2 of variable t, which did is significant and has a positive sign, which suggests that not return a conclusive result. the large share of current expenditure in the composition Nonetheless, the parameters of the stochastic of aggregate spending in Latin American countries, on production frontier estimated in terms of the trend average, produces inefficiency in the economy. To some components, provide clear evidence that technical extent, these results agree with those obtained by Klein progress occurred at an increasing rate (shown by the and Luu (2001), who concluded that countries with high positive sign of the t2 variable, which is significant at levels of current expenditure tend to be less efficient, 1%), which thus means an acceleration in the variation since a high level of public expenditure crowds out of technical progress. productive investments and thus generates distortions The value of the technical inefficiency indicator, in resource allocation. g, is 0.51. This means that 51% of the total variance of The coefficient of the inflation rate (z3t) proved the composed error of the estimation of the translog positive and significant, in keeping with the empirical production function is explained by the variance in literature that shows the harmful effects that higher rates technical inefficiency. This makes it very important to of inflation have on resource allocation in the economy. incorporate technical inefficiency into the model. Such rates end up inhibiting trade and discouraging capital Table 1 shows that all of the estimated parameters of formation. In this context, it is important to stress that the variables included to explain the technical inefficiency several Latin American countries experienced lengthy

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA 58 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 1 Results of the model in the transloga functional form, 1961-2010 Variables Estimations Z-value d1 0.395 8.8 d2 -0.683 -13.5 d3 1.490 9.9 d4 -0.226 -8.2 d5 0.416 10.3 d6 -0.592 -7.9 d7 -0.475 -10.5 d8 -0.770 -24.6 d9 -0.540 -9.3 d10 -0.400 -9.5 d11 -0.930 -14.6 d12 -0.931 -13.6 d13 1.150 14.4 d14 -1.020 -13.0 d15 -1.073 -18.2 d16 -0.155 -5.5 d17 -0.844 -7.5 d18 -0.934 -16.8 t -0.014 -1.3 1 t2 2 0.001 11.8 LnL 0.973 4.4 LnK -0.327 -2.2 tLnL 0.001 1.8 tLnK -0.002 -3.2 LnL ∙ LnK 0.061 5.8 1 2 LnL 2 _ i -0.069 -4.4 1 2 LnK 2 _ i 0.024 2.9 Constants 11.22 11.2 Z1-trend effect 0.018 4.3 Z2-government consumption expenditure 30.729 5.7 Z3-inflation rate 0.098 2.9 Z4-degree of openness -0.806 -3.4 Constants -0.696 -3.3 lnsigma2 -3.829 -17.7 Ilgtgama 0.042 0.93 sigma2 0.021 - gamma 0.510 - sigma_u2 0.011 - sigma_v2 0.010 -

Source: prepared by the authors, on the basis of the research data. a Number of observations: 950; log-likelihood probability: 729.28 and probability>chi-squared = 0.0000. Note: the “d” variables represent the fixed effects of the countries. The other variables correspond to those indicated in equation (2). Gamma and sigma correspond to the results of the log-likelihood function expressed in terms of the parameterization specified 2 vu by c = 22. vvuv+

inflationary periods, which had negative repercussions trade policies based on exchange-rate devaluation on the technical inefficiency and development of succeeded in reducing the level of inefficiency. This type their economies. of devaluation can boost exports and consequently support In the case of the variable that captures the deviation an increase in the installed capacity of the external sector, of local prices from purchasing power parity (z4t), the which leads in turn to an expansion of domestic demand. estimated coefficient is significant and has the expected It is worth highlighting that this effect is greater, the negative sign. This indicates that countries which adopted larger is share of the external sector in the local economy.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 59

2. Analysis of the hypothesis tests the hypothesis that all of the second-order coefficients and the coefficients of the cross-products of the function Once the models had been estimated, the respective defined in (2) are equal to zero. It should be noted that tests were conducted for functional form, the absence the value of the likelihood ratio was 9.77, which is above of technical progress and technical inefficiency. the 7.04 critical level of the value statistic of Kodde Table 2 sets out some of the results. Firstly, the and Palm (1986) (critical value to the right of the c2 Cobb-Douglas functional form was tested in comparison distribution at 5% with 3 degrees of freedom). It can with the translog model, and then the likelihood-ratio was therefore be assumed that the most appropriate model for used to verify the best functional form. This means testing the problem under study is the translog functional form.

TABLE 2 Likelihood-ratio test of the parameters of the stochastic production frontier Decision Test Null hypothesis Critical value Value of l (5% level)

Functional form H0: α8 = α9 = α10 = 0 9.55 7.04 Reject H0 Absence of tp H0: α2 = α3 = α5 = α7 = 0 161.23 8.76 Reject H0 Nonexistence of technical inefficiency H0: z1 = z2 = z3 = z4 = 0 118.01 8.76 Reject H0 Source: prepared by the authors. l: Statistical test of the likelihood ratio in which l = -2 {log [likelihood (H0)] – log [likelihood (H1)]}. This test has a roughly chi-squared distribution with degrees of freedom equal to the number of independent constraints. tp: technical progress.

Once the functional form had been chosen, the country averages of the decomposition throughout the absence of technical progress was tested. In line with period analysed (1962-2010).1 The results shown in the test described above, the model was estimated in tables 3, 4, 5, 6, 7 and 8 are the average values for each the translog functional form, and with the absence of country over 10-year time intervals. technical progress. The respective values of the log- The average economic growth rate in Latin America maximum-likelihood of each estimation were used to in the 50 years of the study was 4.2%, whereas the rate obtain lr = - 2 [648.67 – 729.28] = 161.22. The result of of change of tfp for the sample as a whole was -0.3% the test exceeds the critical value of 8.76 with 4 degrees in that period (see table 3). The following tables present of freedom and significant at 5% in the table of Kodde those rates separately for each country. and Palm (1986). Consequently, H0 is rejected, and In general, the results agree with those obtained in hypothesis H1 is accepted, which confirms the presence the studies by de Fajnzylber, Loayza and Calderón (2002) of technical progress. on the growth of the Latin American economies and Subsequently, the test for the absence of technical other countries. Tables 6 and 7 show that the economic inefficiency was applied to the model, with the following growth rates of some countries in the 1990s were less than results: lr = - 2 [670.28 – 729.28] = 118.01. Nonetheless, those recorded in the previous decade. A case in point the critical value of the Kodde and Palm table is 8.76, is Colombia, which grew by 4.67% between 1981 and with 4 degrees of freedom and a 5% significance interval. 1990, and by 4.55% in the following decade. Cárdenas Consequently, the value of the maximum-likelihood (2007), who obtains similar results, also shows that the ratio exceeds the critical value of the Kodde and Palm long-term economic growth rate in Colombia has fallen (1986) table, thus indicating the presence of technical since 1980 owing to increasing levels of violence fuelled inefficiency in the model. by an expansion in drug trafficking activities. Among other countries, the Bolivarian Republic of Venezuela 3. tfp and its components recorded growth of 3.7% between 1980 and 1990, but only 3.6% in the following decade. Based on the results of the model estimation obtained above and the income distribution data (sK and sL), 1 The breakdown was performed as from 1962 owing to the availability total factor productivity is broken down according to of data on the variation of technical inefficiency based on the translog the model described in section IV. Table 3 shows the model estimated using 1961-2010 data.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA 60 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 3 tfp results: averages 1962-2010 Economic Capital Expansion of Change Technical Technical Scale Distributive Random Country growth accumulation labour force in tfp progress efficiency economies gains disturbances Argentina 0.0360 0.0496 0.0161 0.0002 0.0021 -0.0013 0.0102 -0.0109 -0.0299 Bolivia (Plurinational State of) 0.0400 0.0600 0.0240 -0.0093 -0.0019 -0.0014 0.0087 -0.0147 -0.0347 Brazil 0.0550 0.0604 0.0301 0.0096 0.0044 -0.0004 0.0136 -0.0080 -0.0451 Chile 0.0427 0.0553 0.0230 -0.0011 0.0002 -0.0005 0.0116 -0.0124 -0.0346 Colombia 0.0503 0.0583 0.0309 0.0043 0.0027 -0.0014 0.0125 -0.0096 -0.0432 Costa Rica 0.0516 0.0648 0.0348 -0.0049 -0.0022 -0.0038 0.0151 -0.0140 -0.0431 Dominican Republic 0.0541 0.0838 0.0306 -0.0101 0.0010 -0.0010 0.0126 -0.0226 -0.0502 Ecuador 0.0392 0.0419 0.0288 0.0038 0.0044 -0.0028 0.0056 -0.0034 -0.0354 El Salvador 0.0355 0.0583 0.0233 -0.0101 0.0004 -0.0048 0.0095 -0.0152 -0.0360 Guatemala 0.0411 0.0579 0.0263 -0.0051 0.0007 -0.0041 0.0111 -0.0128 -0.0381 Honduras 0.0448 0.0644 0.0327 -0.0079 0.0006 -0.0062 0.0119 -0.0142 -0.0444 Jamaica 0.0229 0.0501 0.0131 -0.0177 -0.0035 -0.0062 0.0084 -0.0165 -0.0226 Mexico 0.0575 0.0607 0.0333 0.0081 0.0017 -0.0011 0.0160 -0.0085 -0.0446 Nicaragua 0.0409 0.0555 0.0344 -0.0077 -0.0005 -0.0098 0.0125 -0.0099 -0.0412 Paraguay 0.0459 0.0681 0.0291 -0.0071 0.0014 -0.0015 0.0103 -0.0174 -0.0442 Peru 0.0450 0.0416 0.0301 0.0086 0.0011 -0.0006 0.0121 -0.0040 -0.0352 Trinidad and Tobago 0.0327 0.0526 0.0162 -0.0137 -0.0066 -0.0005 0.0106 -0.0173 -0.0223 Uruguay 0.0231 0.0509 0.0078 -0.0145 -0.0018 -0.0005 0.0064 -0.0186 -0.0211 Venezuela (Bolivarian Republic of) 0.0522 0.0424 0.0352 0.0113 -0.0010 -0.0004 0.0155 -0.0028 -0.0367

Source: prepared by the authors. tfp: total factor productivity.

TABLE 4 tfp decomposition: averages 1962-1970 Economic Capital Expansion of Change Technical Technical Scale Distributive Random Country growth accumulation labour force in tfp progress efficiency economies gains disturbances Argentina 0.0802 0.1549 0.0143 -0.0438 -0.0183 -0.0006 0.0205 -0.0454 -0.0452 Bolivia (Plurinational State of) 0.0681 0.1785 0.0182 -0.0717 -0.0174 -0.0012 0.0132 -0.0663 -0.0569 Brazil 0.1014 0.1568 0.0328 -0.0263 -0.0143 -0.0008 0.0215 -0.0328 -0.0619 Chile 0.0641 0.1421 0.0137 -0.0529 -0.0198 -0.0005 0.0161 -0.0487 -0.0388 Colombia 0.0771 0.1477 0.0242 -0.0438 -0.0165 -0.0005 0.0162 -0.0431 -0.0509 Costa Rica 0.0718 0.1599 0.0327 -0.0647 -0.0212 -0.0012 0.0172 -0.0596 -0.0560 Dominican Republic 0.0939 0.2030 0.0417 -0.0695 -0.0169 -0.0007 0.0168 -0.0688 -0.0813 Ecuador 0.0723 0.1450 0.0259 -0.0506 -0.0195 -0.0014 0.0177 -0.0474 -0.0481 El Salvador 0.0812 0.1701 0.0364 -0.0609 -0.0188 -0.0013 0.0174 -0.0581 -0.0644 Guatemala 0.0764 0.1709 0.0256 -0.0619 -0.0182 -0.0012 0.0165 -0.0590 -0.0582 Honduras 0.0759 0.1909 0.0310 -0.0775 -0.0185 -0.0029 0.0155 -0.0717 -0.0685 Jamaica 0.0579 0.1626 0.0114 -0.0745 -0.0227 -0.0008 0.0159 -0.0669 -0.0416 Mexico 0.1066 0.1802 0.0299 -0.0385 -0.0175 -0.0005 0.0263 -0.0468 -0.0650 Nicaragua 0.0826 0.2100 0.0321 -0.0854 -0.0203 -0.0005 0.0183 -0.0830 -0.0741 Paraguay 0.0608 0.1528 0.0260 -0.0640 -0.0163 -0.0007 0.0098 -0.0567 -0.0540 Peru 0.0781 0.1452 0.0235 -0.0440 -0.0195 -0.0005 0.0200 -0.0441 -0.0466 Trinidad and Tobago 0.0427 0.1243 0.0110 -0.0672 -0.0252 -0.0005 0.0125 -0.0540 -0.0254 Uruguay 0.0436 0.1174 0.0083 -0.0561 -0.0205 -0.0007 0.0108 -0.0458 -0.0260 Venezuela (Bolivarian Republic of) 0.0853 0.1485 0.0304 -0.0430 -0.0212 -0.0004 0.0236 -0.0451 -0.0505

Source: prepared by the authors. tfp: total factor productivity.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 61

TABLE 5 tfp decomposition: averages 1971-1980 Economic Capital Expansion of Change Technical Technical Scale Distributive Random Country growth accumulation labour force in tfp progress efficiency economies gains disturbances Argentina 0.0409 0.0608 0.0145 -0.0133 -0.0096 0.0001 0.0117 -0.0156 -0.0211 Bolivia (Plurinational State of) 0.0375 0.0495 0.0228 -0.0108 -0.0081 0.0001 0.0084 -0.0112 -0.0240 Brazil 0.0810 0.1003 0.0359 -0.0026 -0.0059 0.0000 0.0208 -0.0176 -0.0525 Chile 0.0334 0.0192 0.0260 0.0020 -0.0098 0.0002 0.0089 0.0026 -0.0139 Colombia 0.0521 0.0612 0.0302 -0.0056 -0.0074 0.0001 0.0127 -0.0110 -0.0336 Costa Rica 0.0667 0.0837 0.0410 -0.0144 -0.0124 -0.0007 0.0185 -0.0198 -0.0436 Dominican Republic 0.0681 0.1119 0.0345 -0.0255 -0.0088 -0.0001 0.0161 -0.0327 -0.0528 Ecuador 0.0211 0.0353 0.0273 -0.0111 0.0125 -0.0091 -0.0196 0.0051 -0.0304 El Salvador 0.0462 0.0699 0.0299 -0.0194 -0.0098 -0.0049 0.0126 -0.0174 -0.0341 Guatemala 0.0479 0.0762 0.0227 -0.0192 -0.0097 0.0003 0.0122 -0.0220 -0.0318 Honduras 0.0418 0.0678 0.0230 -0.0201 -0.0098 0.0000 0.0101 -0.0203 -0.0289 Jamaica 0.0367 0.0339 0.0299 -0.0096 -0.0138 -0.0074 0.0133 -0.0017 -0.0175 Mexico 0.0781 0.0671 0.0505 0.0084 -0.0084 0.0001 0.0219 -0.0052 -0.0479 Nicaragua 0.0455 0.0459 0.0368 -0.0086 -0.0114 -0.0066 0.0135 -0.0042 -0.0286 Paraguay 0.0619 0.1194 0.0301 -0.0343 -0.0081 0.0006 0.0131 -0.0399 -0.0533 Peru 0.0402 0.0248 0.0312 0.0036 -0.0096 -0.0002 0.0111 0.0023 -0.0194 Trinidad and Tobago 0.0636 0.1078 0.0299 -0.0334 -0.0165 -0.0001 0.0207 -0.0375 -0.0406 Uruguay 0.0256 0.0734 0.0035 -0.0347 -0.0114 0.0002 0.0070 -0.0305 -0.0165 Venezuela (Bolivarian Republic of) 0.0745 0.0644 0.0469 0.0040 -0.0121 0.0001 0.0228 -0.0067 -0.0409

Source: prepared by the authors. tfp: total factor productivity.

TABLE 6 tfp decomposition: averages 1981-1990 Economic Capital Expansion of Change Technical Technical Scale Distributive Random Country growth accumulation labour force in tfp progress efficiency economies gains disturbances Argentina 0.0059 0.0046 0.0180 -0.0021 0.0013 -0.0139 0.0058 0.0047 -0.0147 Bolivia (Plurinational State of) 0.0220 -0.0005 0.0246 0.0168 0.0031 -0.0026 0.0056 0.0108 -0.0189 Brazil 0.0325 0.0222 0.0311 0.0117 0.0046 -0.0058 0.0105 0.0025 -0.0325 Chile 0.0333 0.0128 0.0290 0.0164 0.0017 -0.0007 0.0090 0.0064 -0.0251 Colombia 0.0467 0.0319 0.0369 0.0164 0.0033 -0.0010 0.0123 0.0018 -0.0384 Costa Rica 0.0363 0.0214 0.0369 0.0079 -0.0016 -0.0113 0.0136 0.0072 -0.0299 Dominican Republic 0.0401 0.0327 0.0317 0.0092 0.0015 -0.0028 0.0111 -0.0004 -0.0335 Ecuador 0.0355 0.0140 0.0312 0.0155 0.0000 -0.0024 0.0109 0.0070 -0.0252 El Salvador 0.0121 0.0046 0.0166 0.0045 0.0011 -0.0068 0.0048 0.0053 -0.0135 Guatemala 0.0280 0.0074 0.0298 0.0139 0.0011 -0.0055 0.0089 0.0094 -0.0231 Honduras 0.0395 0.0116 0.0386 0.0196 0.0011 -0.0048 0.0110 0.0122 -0.0303 Jamaica 0.0178 0.0152 0.0104 0.0019 -0.0024 0.0017 0.0048 -0.0023 -0.0097 Mexico 0.0217 0.0206 0.0136 0.0055 0.0019 -0.0009 0.0067 -0.0022 -0.0180 Nicaragua 0.0033 0.0204 0.0339 -0.0220 -0.0003 -0.0394 0.0113 0.0063 -0.0289 Paraguay 0.0501 0.0541 0.0348 0.0041 0.0013 -0.0014 0.0129 -0.0086 -0.0428 Peru 0.0239 0.0042 0.0318 0.0119 0.0019 -0.0094 0.0091 0.0103 -0.0241 Trinidad and Tobago -0.0017 0.0196 0.0093 -0.0244 -0.0067 -0.0186 0.0058 -0.0049 -0.0062 Uruguay 0.0110 0.0121 0.0082 -0.0010 -0.0012 -0.0017 0.0037 -0.0017 -0.0083 Venezuela (Bolivarian Republic of) 0.0379 -0.0056 0.0401 0.0268 -0.0007 -0.0023 0.0124 0.0174 -0.0234

Source: prepared by the authors. tfp: total factor productivity.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA 62 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 7 tfp decomposition: averages 1991-2000 Economic Capital Expansion of Change Technical Technical Scale Distributive Random Country growth accumulation labour force in tfp progress efficiency economies gains disturbances Argentina 0.0327 0.0128 0.0144 0.0322 0.0129 0.0131 0.0057 0.0006 -0.0267 Bolivia (Plurinational State of) 0.0323 0.0124 0.0305 0.0285 0.0149 -0.0018 0.0075 0.0079 -0.0391 Brazil 0.0344 0.0124 0.0292 0.0322 0.0161 0.0028 0.0086 0.0047 -0.0395 Chile 0.0372 0.0532 0.0184 0.0087 0.0125 0.0001 0.0102 -0.0140 -0.0431 Colombia 0.0455 0.0263 0.0406 0.0287 0.0145 -0.0039 0.0130 0.0051 -0.0501 Costa Rica 0.0460 0.0235 0.0335 0.0287 0.0097 0.0015 0.0129 0.0046 -0.0397 Dominican Republic 0.0357 0.0407 0.0218 0.0140 0.0124 0.0004 0.0093 -0.0082 -0.0408 Ecuador 0.0327 -0.0001 0.0338 0.0328 0.0118 -0.0030 0.0100 0.0139 -0.0337 El Salvador 0.0316 0.0255 0.0188 0.0209 0.0125 0.0047 0.0068 -0.0031 -0.0336 Guatemala 0.0247 0.0166 0.0217 0.0190 0.0127 -0.0031 0.0073 0.0021 -0.0326 Honduras 0.0418 0.0336 0.0441 0.0170 0.0125 -0.0145 0.0142 0.0048 -0.0530 Jamaica 0.0065 0.0259 0.0069 -0.0043 0.0085 -0.0084 0.0045 -0.0089 -0.0220 Mexico 0.0562 0.0194 0.0503 0.0398 0.0138 -0.0003 0.0165 0.0098 -0.0533 Nicaragua 0.0515 -0.0018 0.0397 0.0507 0.0119 0.0084 0.0111 0.0194 -0.0371 Paraguay 0.0297 0.0181 0.0267 0.0210 0.0123 -0.0039 0.0086 0.0039 -0.0360 Peru 0.0480 0.0101 0.0376 0.0421 0.0138 0.0071 0.0109 0.0103 -0.0418 Trinidad and Tobago 0.0331 -0.0066 0.0188 0.0360 0.0052 0.0107 0.0075 0.0126 -0.0150 Uruguay 0.0205 0.0300 0.0104 0.0072 0.0099 0.0008 0.0057 -0.0091 -0.0271 Venezuela (Bolivarian Republic of) 0.0369 -0.0047 0.0361 0.0388 0.0116 0.0010 0.0107 0.0156 -0.0334

Source: prepared by the authors. tfp: total factor productivity.

TABLE 8 tfp decomposition: averages 2001-2010 Economic Capital Expansion of Change Technical Technical Scale Distributive Random Country growth accumulation labour force in tfp progress efficiency economies gains disturbances Argentina 0.0201 0.0150 0.0192 0.0279 0.0243 -0.0050 0.0072 0.0014 -0.0420 Bolivia (Plurinational State of) 0.0254 0.0063 0.0281 0.0380 0.0265 -0.0047 0.0065 0.0096 -0.0471 Brazil 0.0257 0.0101 0.0217 0.0328 0.0213 0.0017 0.0065 0.0033 -0.0390 Chile 0.0455 0.0492 0.0279 0.0204 0.0166 -0.0015 0.0138 -0.0085 -0.0520 Colombia 0.0300 0.0246 0.0229 0.0257 0.0196 -0.0019 0.0085 -0.0006 -0.0431 Costa Rica 0.0370 0.0354 0.0300 0.0179 0.0145 -0.0073 0.0131 -0.0024 -0.0464 Dominican Republic 0.0326 0.0306 0.0233 0.0213 0.0168 -0.0019 0.0096 -0.0031 -0.0426 Ecuador 0.0342 0.0152 0.0261 0.0326 0.0174 0.0017 0.0091 0.0044 -0.0396 El Salvador 0.0062 0.0212 0.0150 0.0043 0.0171 -0.0157 0.0058 -0.0028 -0.0343 Guatemala 0.0282 0.0184 0.0317 0.0228 0.0177 -0.0108 0.0104 0.0055 -0.0448 Honduras 0.0250 0.0179 0.0268 0.0214 0.0175 -0.0087 0.0088 0.0039 -0.0412 Jamaica -0.0045 0.0127 0.0070 -0.0019 0.0132 -0.0160 0.0036 -0.0027 -0.0222 Mexico 0.0248 0.0164 0.0220 0.0252 0.0187 -0.0038 0.0085 0.0017 -0.0387 Nicaragua 0.0216 0.0030 0.0293 0.0268 0.0174 -0.0110 0.0082 0.0121 -0.0375 Paraguay 0.0268 -0.0041 0.0281 0.0375 0.0179 -0.0020 0.0073 0.0144 -0.0347 Peru 0.0347 0.0235 0.0262 0.0293 0.0191 -0.0002 0.0094 0.0011 -0.0443 Trinidad and Tobago 0.0258 0.0178 0.0122 0.0202 0.0103 0.0061 0.0066 -0.0028 -0.0244 Uruguay 0.0147 0.0218 0.0085 0.0120 0.0145 -0.0010 0.0046 -0.0061 -0.0277 Venezuela (Bolivarian Republic of) 0.0266 0.0094 0.0226 0.0298 0.0172 -0.0004 0.0079 0.0050 -0.0352

Source: prepared by the authors. tfp: total factor productivity.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 63

Table 3 also shows that the countries that recorded a Tables 4 and 5 show that Brazil suffered allocative larger contribution of technical progress to productivity efficiency losses in the first two decades analysed, which growth in the 50-year period analysed were Argentina, are the clear result of a growth strategy that did not take Brazil, Colombia and Ecuador, with indices of around account of the adjustment. It can also be seen that both 0.3%. Brazil displayed an average index of 0.4%, as well output and physical capital grew more strongly in the as the highest indices in the last three decades (see tables 1970s than in other decades. Those findings agree with 6, 7 and 8). These results coincide with those reported those obtained by Pires and Garcia (2004), who argue that by Pires and Garcia (2004), which also identified low in the 1970s there was an intensive process of resource rates of technical progress in Brazil between 1970 and allocation in the economy, which led to a considerable 2000, because the authors take account of the fact that investment in infrastructure in Brazil. this country was not a member of the Organisation The analysis of the data presented in tables 4 and for Economic Co-operation and Development (oecd), 5 shows that the indices of economic growth in Brazil and that the markets of the Bolivarian Republic of were higher in the first two decades before dropping to Venezuela, Mexico and Peru underwent a process of around 3% between 1980 and 2000. This is explained by a import substitution related to episodes of economic slackening of growth in the country owing to the exhaustion liberalization, during which the industrialization process of the industrialization-via-import-substitution model. slowed down. In the five decades examined separately, only As shown in table 3, the 19 countries analysed Trinidad and Tobago posted negative growth in the in this study recorded decreasing technical efficiency, decade 1981-1990. In general, the pattern of economic which assumes that the contribution of that efficiency growth in the countries is similar, and, as shown in table to tfp was negative in all countries. Nonetheless, there 6, the average does not exceed 6% in the period studied. was some technical progress in most cases, and output The cases of Brazil, Colombia and Paraguay stand out increased in all of them. It is well known that technical as those with the highest indices of economic growth, efficiency is determined by the distance from the averaging 4.3%. The countries with the lowest average technological frontier and effective use of technologies, growth indices were Argentina and Uruguay, with just so these results suggest that the expansion of the frontier 0.59% and 1.10%, respectively. was more intense and faster than the dissemination of As can be seen in table 4, all of the Latin American new technologies. In other words, some of the countries countries analysed displayed negative indices of tfp analysed did not fully keep pace with the technological variation in the first period studied (1962-1970). This developments that occurred in the period analysed, situation changed in the subsequent decades, when some possibly owing to problems in the process of diffusion countries displayed positive indices. Table 7 and 8 show and adoption of more modern technologies. that all countries except the Jamaica achieved positive In a general analysis of the decomposition of tfp, indices of tfp growth in the decades of 1990 and 2000. The table 8 shows that most of the countries display positive average growth rate of Brazilian productivity throughout allocative gains, including Brazil. Those results reflect the period analysed were 0.9% per year (see table 3). improvements in resource allocation among the factors Table 8 shows that in the analysis of the decade of of production used in those countries. 2000, there was positive economic growth (averaging These results agree with those reported by Pires 2.5%) in all countries except Jamaica, where output and Garcia (2004), whose estimations show that Costa declined by about 0.04%, with Chile growing by an Rica and Trinidad and Tobago displayed the largest average of 4.5%. Technical progress and economies distributive gains, represented by indices of 4.2% and of scale were positive in all countries; but, while some 13.5%, respectively. Those two countries were also the recorded positive indices with respect to distributive gains, leaders in the sample in terms of technical progress such as Brazil, the equivalent indices were negative in during the period studied. others, such as Uruguay.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA 64 CEPAL REVIEW 114 • DECEMBER 2014

VI Final thoughts

The analysis of tfp and its components in Latin America countries studied was positive. Brazil is one of the leading in the period 1960-2010, using a stochastic-frontier model countries in this respect, with a growth rate of 5.5%. The which includes macroeconomic variables of technical analysis of the 1960s and 1970s shows that the average inefficiency, shows that those variables generally have Brazilian growth rates were around 7%, which possibly a significant effect that enables better understanding of coincides with the adoption of the import substitution technical inefficiency throughout the region. industrialization model in the countries of the region. The significance of the effects is found both Costa Rica, the Dominican Republic, Ecuador, through likelihood tests and through the parameter γ, of Guatemala, Mexico and Paraguay recorded similar value 0.51, in the model estimation. average gdp growth rates of 5.1%; 4.0%; 4.1%; 5.7%; The most important variables for explaining the 4.5%, and 5.4%, respectively. The worst performer in technical inefficiency of the sample countries, in other the period was Uruguay, where the average growth rate words those that display a positive relation to inefficiency, was just 2.3%. are public expenditure and the inflation rate: the higher The results of the decomposition of the change in tfp these rates are, the more they are associated with technical into technical progress, technical efficiency, economies inefficiency. of scale and distributive gains vary between the countries In contrast, the variable corresponding to the analysed. Although there is unanimity with respect to deviation of local prices from purchasing power parity technical progress (the average was positive in most (used as a proxy variable for the exchange rate) displays of the countries throughout the period analysed), the an inverse relation with respect to technical inefficiency: results in terms of the other components are different. the larger deviation of this relative price, the less is Lastly, it is worth stressing that the great advantage technical inefficiency. of this tfp decomposition model compared to that known Although relatively low throughout the period as the Malmquist index is the possibility of incorporating studied, the average rate of economic growth of the scale and allocative effects in the analysis of the results.

Bibliography

Aigner, D.J., C.A.K. Lovell and P. Schmidt (1977), “Formulation and Coelli, T.J., D.S.P. Rao and G.E. Battese (1998), An Introduction estimation of stochastic frontier production functions models”, to Efficiency and Productivity Analysis, Kluwer Academic Journal of Econometrics, vol. 6, No. 1, Amsterdam, Elsevier. Publishers. Battese, G.E. and T.J. Coelli (1995), “A model for technical De Gregorio, J. (1992), “Economic growth in Latin America”, Journal inefficiency effects in stochastic frontier production functions of Development Economics, vol. 39, No. 1, Amsterdam, Elsevier. for panel data”, Empirical Economics, vol. 20, No. 2, Springer. Fajnzylber, P., N. Loayza and C. Calderón (2002), Economic (1993), “A stochastic frontier production incorporating Growth in Latin America and the Caribbean: Stylized Facts, a model for technical inefficiency effects”, Working Papers Explanation and Forecasts, Washington, D.C., World Bank. in Econometrics and Applied Statistics, No. 69, Armidale, Färe, R. and others (1994), “Productivity growth, technical progress, University of New England. and efficiency change in industrialized countries”, American Battese, G.E. and G.S. Corra (1977), “Estimation of a production Economic Review, vol. 84, No. 1, Nashville, Tennessee, frontier model: with application to the pastoral zone of eastern American Economic Association. Australia”, Australian Journal of Agricultural Economics, (1992), “Productivity changes in Swedish pharmacies vol. 21, No. 3, Canberra, Australian Agricultural and Resource 1980-1989: a non-parametric Malmquist approach”, Journal Economics Society. of Productivity Analysis, vol. 3, No. 1-2, Springer. Bauer, P.W. (1990), “Recent developments in the econometric Färe, R., S. Grosskopf and P. Roos (1998), “Malmquist productivity estimation of frontiers”, Journal of Econometrics, vol. 46, indexes: a survey of theory and practice”, Index Numbers: Essays No. 1-2, Amsterdam, Elsevier. in Honour of Sten Malmquist, Kluwer Academic Publishers. Bittencourt, A. and Marinho, E. (2007), “Produtividade e crescimento Griliches, Z. (1996), “The discovery of the residual: a historical note”, econômico na América Latina: a abordagem da fronteira de Journal of Economic Literature, vol. 34, No. 3, Nashville, produção estocástica”, Estudos Econômicos, vol. 37, No. 1, Tennessee, American Economic Association. São Paulo, Instituto de Pesquisas Econômicas. Henningsen, A. and C.H. Henning (2009), “Imposing regional Cárdenas, M. (2007), “Economic growth in Colombia: a reversal of monotonicity on translog stochastic production frontiers with a fortune?”, Ensayos sobre Política Económica, vol. 25, No. 53, simple three-step procedure”, Journal of Productivity Analysis, Bogota, Bank of the Republic. vol. 32, No. 3, Springer.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA CEPAL REVIEW 114 • DECEMBER 2014 65

Johnson, A.L. and T. Kuosmanen (2012), “One-stage and two- Meeusen, W. and V.D. Broeck (1977), “Efficiency estimation from stage dea estimation of the effects of contextual variable”, Cobb-Douglas production functions with composed error”, European Journal of Operational Research, vol. 220, No. 2, International Economic Review, vol. 18, No. 2, Wiley. Amsterdam, Elsevier. Orea, L. (2002), “Parametric decomposition of a generalized Klein, P.G. and H. Luu (2001), Politics and Productivity, Merril Malmquist productivity index”, Journal of Productivity Lynch Capital Markets Bank Ltd. Analysis, vol. 18, No. 1, Kluwer Academic Publishers. Kodde, D.A. and F.C. Palm (1986), “Wald criteria for jointly testing Pires, J.O. and F. Garcia (2004), “Productivity of nations: a stochastic equality and inequality restrictions”, Econometrica, vol. 54, frontier approach to tfp decomposition”, Econometric Society No. 5, New York, The Econometric Society. Latin American Meetings, No. 292, Econometric Society. Kumbhakar, S.C. (2000), “Estimation and decomposition of Pitt, M.M. and L.F. Lee (1981), “The measurement and sources productivity change when production is not efficient”, of technical inefficiency in the Indonesian weaving Econometric Reviews, vol. 19, No. 4, Taylor & Francis. industry”, Journal of Development Economics, vol. 9, No. 1, Kumbhakar, S.C. and C.A.K. Lovell (2003), Stochastic Frontier Amsterdam, Elsevier. Analysis, New York, Cambridge University Press. Sauer, J., K. Frohberg and H. Hockmann (2006), “Stochastic Laborda, L., D. Sotelsek and J.L. Guasch (2011), “Innovative and efficiency measurement: the curse of theoretical consistency”, absorptive capacity of international knowledge: an empirical Journal of Applied Economics, vol. 9, No. 1, cema University. analysis of productivity sources in Latin American countries”, Schmidt, P. and R. Sickles (1984), “Production frontiers and panel Latin American Business Review, vol. 12, No. 4, Taylor data”, Journal of Business and Economic Statistics, vol. 2, & Francis. No. 4, Taylor & Francis. Lee, C.Y. and others (2013), “A more efficient algorithm for convex Solow, R.M. (1957), “Technical change and the aggregate production nonparametric least squares”, European Journal of Operational function”, Review of Economic and Statistics, vol. 39, No. 3, Research, vol. 227, No. 2, Amsterdam, Elsevier. Cambridge, Massachusetts, The mit Press. Maudos, J., J.M. Pastor and L. Serrano (1999), “Total factor Wang, Y. and others (2014), “Nonparametric quantile frontier productivity measurement and human capital in oecd countries”, estimation under shape restriction”, European Journal of Economics Letters, vol. 63, No. 1, Amsterdam, Elsevier. Operational Research, vol. 232, No. 2, Amsterdam, Elsevier.

LATIN AMERICA: TOTAL FACTOR PRODUCTIVITY AND ITS COMPONENTS • JAIR ANDRADE ARAUJO, DÉBORA GASPAR FEITOSA AND ALMIR BITTENCOURT DA SILVA

Financial constraints on economic development: Theory and policy for developing countries

Jennifer Hermann

ABSTRACT This article contains a theoretical and policy analysis of the financial constraints on economic development in developing countries. Following a Keynesian interpretation, it concludes that financial policies are needed to relieve these constraints, given the natural tendency of financial systems to operate in ways that are dysfunctional to economic development. It then proposes three lines of policy that take account of the special characteristics of developing countries: resource allocation policies targeted at segments of strategic importance for economic and financial development; policies to control financial and external fragility; and compensatory policies of a more interventionist cast, in particular directed credit programmes for both public- and private-sector lending to complement resource allocation policies, and countercyclical regulatory barriers so that fragility can be better controlled.

KEYWORDS Economic development, development finance, financial policy, financial services, credit, Keynesian economics, developing countries

JEL CLASSIFICATION G18, O16, O50

AUTHOR Jennifer Hermann is a Professor at the Institute of Economics of the Federal University of Rio de Janeiro (ie/ufrj), Brazil. [email protected] 68 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

The importance of financial dynamism to economic systems, which are usually limited and undiversified development is generally acknowledged and has been (Gerschenkron, 1962; Furtado, 1967). In these cases, empirically confirmed in innumerable studies on the economic performance may be directly constrained by subject. History shows, however, that there is no one the shortcomings of a country’s financial system. ideal model for a financial system capable of supporting The present article is a contribution to this debate economic development and that very few countries have and contains a theoretical and policy analysis intended managed to combine financial and economic development. to assist in the task of understanding and identifying The only instances that can be cited are Germany, the financial constraints on economic development in United Kingdom and the United States, the latter two developing countries. At the theoretical level, it offers with a financial system based on capital markets, the a Keynesian interpretation drawing on the work of two former with one based on private credit (Zysman, 1983). offshoots of this tradition, the new-Keynesian school, Other countries whose industrialization accelerated in represented here by the contributions of J. Stiglitz (Stiglitz the post-war period, including the developing countries and Weiss, 1981; Stiglitz, 1994), and post-Keynesian studied in this article, backed the process with some thinking as derived from the writings of J.M. Keynes combination of public and external credit, and the varying (1943, 1937a and 1937b) and H. Minsky (1982 and degrees of success they achieved revealed the mismatch 1986). Setting out from this interpretation, the article between the dynamics of their financial systems and the proposes a basic list of financial policies designed to financing needs of economic development. make the financial system more functional to economic Although generalizing is difficult, it is possible development, bearing in mind the special characteristics to identify features common to countries at similar of the countries aspiring to this. levels of development. In industrialized countries, it is Section II starts the discussion with a summary innovation that is hardest to finance, both in the research of the theoretical debate on the role of the financial and development (r&d) phase and in the new process system in economic development. The sections that implementation phase (Hall, 2002; Matouk, 2010), follow apply the Keynesian approach to the analysis of given the particular riskiness of these investments. developing countries, concentrating on three aspects: In developing countries, the challenges raised by their specific economic development financing needs underdevelopment itself and by competition from more (section III), the conditions under which the financial advanced economies create further difficulties. A number system can contribute to this process (section IV) and of sectors have a particularly acute need for resources, the profile of the financial policy required to expand it but underdevelopment is also a feature of financial (section V). Section VI offers conclusions. II The financial system and economic development

J. Schumpeter (1934) and J.M. Keynes (1943, 1937a and The debate was stoked in the 1950s and afterwards 1937b) laid the foundations for the theoretical debate by the publication of a well-known article by Gurley and about the macroeconomic role of the financial system. Shaw (1955), followed by other important contributions Schumpeter highlighted the importance of credit to that include Gurley and Shaw (1960), Shaw (1973), economic development and Keynes argued that the money McKinnon (1973), Stiglitz (1994) and Minsky (1982 market was the main driver of economic dynamics in and 1986), among others. The first of the articles cited both the short and the long run. gave rise to the Gurley-Shaw model and the next two

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 69 to the Shaw-McKinnon model, both of which derive risk) and supply (by reducing costs) and foster market from the neoclassical theoretical tradition and are briefly diversification (by doing away with regulatory barriers), summarized in the following subsection. The last two, thus promoting financial development. which are described in more detail further on (since they provide the basis for the policies proposed in this 2. The Keynesian approach paper), provided the main theoretical underpinnings for the Keynesian approach in its new-Keynesian and The Keynesian approach to the financial system is post-Keynesian variants, respectively. based on a position that is critical of the efficient market hypothesis. In new-Keynesian theory, this hypothesis is 1. A brief summary of the neoclassical view replaced by that of market failures, structural conditions that, even in free markets, prevent prices from adjusting The Gurley-Shaw model postulates that a diversified to economic fundamentals. In the financial market, financial system providing a variety of ways to allocate the main failure identified is information asymmetry savings is favourable to economic growth. A kind of between potential borrowers and lenders.1 Because of Say’s Law is proposed for the financial market: a supply this asymmetry, banks cannot properly distinguish the of profitable assets, with varied characteristics, tends to differing risk levels of potential borrowers and thus attract a portion of aggregate saving, thus creating its own set efficient interest rates for each project. Given this demand. Thus, the demand for money and equilibrium limitation, they set homogeneous rates for projects interest rates are lowered and the supply of funds to that differ in riskiness. This results in a first source of finance investments is increased. This approach, however, inefficiency: equilibrium interest rates in the market says nothing about what conditions and policies favour for assets do not properly reflect their microeconomic financial development, an issue that is not dealt with fundamentals, because the lender does not perfectly until the Shaw-McKinnon model. know them. Since it is not possible to identify each The Shaw-McKinnon model relies on three core risk, banks follow two conventional evaluation criteria: hypotheses (Fry, 1995; Hermann, 2003): (i) saving is (i) the greater the rate of return, the higher the risk (Tobin, required to finance economic growth; (ii) aggregate 1958), and (ii) the interest rate accepted by borrowers is saving is a positive function of the real interest rate, and indicative of the returns they expect and their appetite (iii) freely operating financial markets ensure that the for risk. Accordingly, even if lenders anticipated high real interest rate, aggregate saving and, by extension, levels of risk, incorporating them fully into interest economic growth settle at their “optimum” levels. rates would not be a solution, owing to the adverse The first two hypotheses are simply applications selection effect (attracting borrowers with a greater risk of well-known neoclassical monetary theory. The third appetite and putting off more conservative ones). Thus, synthesizes the efficient market hypothesis (Malkiel, 1994), potential lenders tend to protect themselves by rationing according to which, in the absence of exogenous barriers credit and causing the market to become incomplete to agents’ free choices, interest rates correctly reflect in segments where it is particularly hard to evaluate or the microeconomic and macroeconomic fundamentals compensate for risk.2 Rationing and incompleteness of the assets concerned. arise most frequently: According to this approach, any market segment (i) In the capital market, as it is highly risky for non- becomes viable if it is free to price assets in accordance professional investors. with their fundamentals. A situation where financing for (ii) In long-term financing and the financing of economic development is constrained is interpreted as a innovations generally, as risks are hard to predict. symptom of malfunctioning not in the financial system, (iii) In lending to small and medium-sized enterprises but in the wider economy of which it is part. (smes). What is recommended for the creation of a “complete” financial market, then, are: (i) liberalization policies so that regulatory obstacles do not prevent prices 1 Other market failures may be noted, such as transaction costs, from adjusting to asset risks and returns; (ii) supervision information costs, positive and negative externalities, incomplete of the sector (to safeguard individual investors), and markets and imperfect competition (Stiglitz, 1994; idb, 2005, (iii) macroeconomic policies focused on monetary stability chap. 11). All except the last are intrinsic to financial markets. 2 A market is deemed incomplete when one or more segments that to keep risk levels low. According to this approach, are theoretically possible do not exist in practice because of a lack of such policies stimulate asset demand (by reducing interest on the demand or supply side.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 70 CEPAL REVIEW 114 • DECEMBER 2014

(iv) In lending to people on low incomes, who are for financing costs: the interest rates demanded for long- considered risky because of their lower levels of term securities by the savers and financial institutions income and collateral, so that in many countries that might acquire them are generally higher than those they are affected by financial exclusion. for short-term securities. In other words, the yield curve The indisputable importance of these segments for (the interest-rate term structure) is typically rising economic development highlights another inefficiency (Cargill, 1983; Hermann, 2011b). This situation makes in the financial system: its tendency to operate in a way the second phase more fraught than the finance phase, that is not functional to this process. implying that activities requiring long-term funding will The post-Keynesian approach is even more radical be penalized more heavily. in its criticism of the efficient market hypothesis, In the finance phase, the crucial actors are commercial its argument being that the main obstacle to market banks; as they take in sight deposits (very short-term efficiency is the uncertainty surrounding all economic liabilities), they also concentrate their assets in short-term decision-making, and particularly financial operations operations. Once this demand has been met, investment (Carvalho, 2010; Hermann, 2011a). As Kregel (1980) generates new income and thence more saving. Long- observed, the problem is not confined to information term funding is not guaranteed even if savings are costs and access, but includes the very existence, at identical in value to new investments (Keynes, 1937a), the critical juncture, of the forward-looking indicators since where there is a strong preference for liquidity, needed to estimate asset risks and returns. Uncertainty, savings will be allocated predominantly to short-term unlike risk, cannot be priced, i.e. cannot be factored into assets (including cash). In this phase, non-bank financial market interest rates. This makes it reasonable that there institutions need to be willing and able to attract new should normally be some preference for liquidity, as a savings for medium- and long-term assets. way of protecting against unforeseeable risks. With the Keynesian approach, in summary, the vital In Keynes’s view, the preference for liquidity is the actors in the investment financing process are not savers main source of macroeconomic inefficiency in resource (firms and families keeping resources in the financial allocation, both in asset markets, where it creates a system) but financial institutions: commercial banks in the short-termist tendency that makes it harder (and more finance phase and other financial institutions (including expensive) to finance investment, and in the market universal banks) in the funding phase (Carvalho, 1997). for goods, where it is the main cause of recessions. This interpretation justifies the central role attributed in This exacerbates the difficulties involved in financing the post-Keynesian approach to the financial system as economic development, as already discussed. The a key factor in economic development. post-Keynesian view is that these difficulties reflect, Besides the preference for liquidity, uncertainty on a larger scale, the difficulties of financing aggregate justifies what Keynes (1943, chap. 12) called conventional investment. According to Keynes, such financing is behaviour: going along with the majority at times when carried out in two stages. The first, which he called the risks are hardest to evaluate. Because of this, the market finance stage, consists in demand for credit from firms tends to amplify asset appreciation or depreciation trends, to initiate new investments, and can be met with short- creating speculative bubbles in the first case and asset term loans. For investments to be completed, however, deflation crises in the second. Minsky (1982 and 1986) firms must be able to issue longer-dated securities adds to this analysis the financial fragility hypothesis, (including shares) whose maturities are compatible with pointing out that: (i) phases of economic growth are always that of the asset to be financed. Keynes called this the accompanied by higher debt; (ii) the level of debt tends funding stage. to rise with the degree of development of the financial As Kregel (1986) observed, the finance stage is not system; (iii) the expectations driving this borrowing obligatory, since investment can be financed directly are surrounded by uncertainty; (iv) the ability to make by issuing shares or long-term debt. This situation is repayments may be jeopardized when these expectations less likely than the one depicted by Keynes, however. are not met, creating the risk of liquidity crises in the In both stages, what makes the investment viable is a financial system, and (v) fulfilment of these expectations context of low (or falling) preference for liquidity (which does not in itself guarantee safety for the market either, reduces equilibrium interest rates in the asset market), as it tends to prolong phases of bullishness, sometimes but willingness to forego liquidity obviously needs to unduly, thus creating speculative bubbles that likewise be greater in the funding phase. This has implications trigger financial crises when they burst (Kregel, 1997).

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 71

Financial fragility in market economies thus has in the country, with progress in this tending to expand structural causes: the uncertainty inherent in financial agents’ access to external financing sources. As for the relationships and the development of the financial latter, a large proportion of speculative borrowing is not system itself. The degree of fragility in the economy an anomaly but a common feature reflecting the chronic in each period will depend on the conditions of the difficulty of securing long-term financing in market lending contracts signed. On this subject, Minsky economies, which induces firms to borrow short in order offers a taxonomy of borrowers, classifying them as: to finance assets with long maturities. Financial system (i) hedgers, whose expected incomes exceed their debt fragility is thus intrinsic to financial development. repayments; (ii) speculative borrowers, whose expected In short, from a post-Keynesian perspective, income (in the early phase of the contract) only covers the financial system tends to operate in a way that is interest costs, making it necessary to refinance the doubly dysfunctional to economic development: market principal, and (iii) Ponzi schemes, used by speculative incompleteness in the segments with the greatest uncertainty borrowers who are even greater risk-takers and expect compounds the tendency towards financial fragility. to refinance the principal and the interest in that initial Keynesian-inspired approaches argue for a number phase. Speculative borrowers (including those engaged in of forms of State action in the financial system. Besides Ponzi schemes) operate with a greater degree of fragility the supervision of the sector suggested in the neoclassical since, besides the risk of their earnings expectations approach, recommended policies include prudential not being met, they depend on new loans to cover the regulation (restrictions on the free allocation of resources) maturity mismatch between their assets (longer) and their ranging from incentives to specific allocations to priority liabilities (shorter). sectors and, given the difficulties of development financing, Thus, financial fragility in the economy is increased more interventionist types of action as well, involving by two factors: the size of agents’ debts and the proportion public-sector and directed private-sector lending policies, of speculative borrowers. The former reflects the rate of with the former perhaps extending to the creation of economic growth and the degree of financial development public-sector development banks.

III The financing of economic development

1. Characteristics of the development process various schools of theory dealing with the issue recognize. The complexity of the development process, however, has Economic development is a complex phenomenon major implications when it comes to understanding and requiring prolonged gross domestic product (gdp) determining financing needs (Studart, 2005; Carvalho, growth in conjunction with structural transformations 2010; Hermann, 2011a). The greatest difficulties include: in the economy. This process is brought about by a (i) The large volume of capital required: this restricts series of investments in new production capacity, in two many firms’ access to what prove to be scarce or ways: the capital stock is expanded and new types of costly outside resources. physical and human capital are brought into use. Thus, (ii) Innovations: new sectors, products and production development always involves some degree of innovation, processes do not have a track record of profitability, which comes about through the introduction of new and this exposes innovative firms to credit rationing, products, sectors of activity, production processes and as well as limiting their ability to self-finance. consumption patterns (Schumpeter, 1934). (iii) Externalities: some of the investments needed to A general movement towards increased investment advance the development process are used to expand of the kind that characterizes each phase of economic productive and urban infrastructure, often with the development needs something to supplement self- characteristics of a public good (with a social return financing. From this basic condition arises the importance greater than the expected microeconomic return), of financial development for economic development, as the which limits private-sector interest.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 72 CEPAL REVIEW 114 • DECEMBER 2014

(iv) Consumption: for investment and innovation to (i) developed countries (also known as industrialized expand sustainably, domestic consumption needs countries because the industrial sector dominated to grow too. As with investment, this increased this process until the late twentieth century), which consumption can hardly be financed from personal head the ranking; resources alone, even if these tend to grow because (ii) underdeveloped countries, a term that gradually fell of increased employment and, possibly, some into disuse so that in the 1980s they became known improvement in the income distribution profile. instead as developing or less developed countries, Thus, favourable borrowing conditions for families with a subdivision by income levels (medium are an important complement to the financing and low); structure of economic development. (iii) emerging economies, a term used for developing (v) Possible effects on the trade balance: some of the countries that implemented trade and financial forward and backward linkage effects created by liberalization policies in the 1980s or 1990s, so the new investment can result in certain sectors that international connections became substantially quickly coming up against the limit of their capacity, more important in their development profile. creating bottlenecks on the supply side, while other The first indicator normally offered as a marker effects may stimulate demand in sectors that have for developing countries is their level of gdp and per the capacity to expand supply but are not very capita income, which have historically been lower than competitive externally. In both cases, imports (of in industrialized countries with similar populations capital, intermediate or consumption goods) will be and land areas. Strictly speaking, what distinguishes strongly stimulated, resulting in balance-of-payments developing countries is that they have remained in this disequilibria and possibly inflation (Tavares, 1979; position for decades or even centuries. Despite this, Furtado, 1967). In this case, the development process developing countries are not exactly characterized by will only be sustainable under favourable external difficulty in achieving economic growth, even if this financing conditions. is a recurrent problem in many of them. The greater (vi) Financial market incompleteness: when the financing difficulty is to generate development, i.e. to associate of a particular firm is considered, it is implicitly growth with structural shifts that enhance economic or explicitly assumed that there is a diversified and social indicators and the country’s position in the financial infrastructure enabling the firm to assess international market: market opportunities and opt for the combination “Underdeveloped economies can experience of its own and outside capital that best suits it. In prolonged phases of growth in their overall and per other words, markets are assumed to be complete. capita output without any diminution of the external According to the theoretical approach adopted in dependency and structural heterogeneity that are this study, as has been shown, this condition is not their essential characteristics” (Furtado, 1967). guaranteed even for individual firms, let alone the What distinguishes developing countries more whole economy. than their relatively low incomes is the difficulty they Given these conditions, it is clear that, by contrast have in implementing and, most particularly, generating with what conventional approaches recommend, the innovations in an international market already dominated financing of economic development requires far more by technologically more advanced countries. Thus, many than free markets and business expertise if “optimum” authors consider technological underdevelopment the capital structures are to be created. In the absence of a main distinguishing feature of developing countries. diversified financial system geared in some degree towards Although hard to measure, this can be clearly perceived dealing with the specific needs of this process, even the in some of its effects: (i) the low productivity of capital most promising projects can be rendered unviable by and labour by developed-country standards; (ii) chronic financing difficulties. difficulty in competing in domestic and external markets even under favourable exchange-rate conditions; 2. Special features of developing countries (iii) import and export patterns concentrated, respectively, in industrial and primary goods; (iv) frequent trade deficits, The effort to understand the heterogeneity of economic and (v) high external indebtedness, among other things. development in different countries gave rise to a Other features of developing countries are a high taxonomy ranking them into three major categories in degree of financial system incompleteness and external the international economy: constraints, although these conditions are also present

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 73 to some degree in industrialized countries.3 The special potential borrowers (Hermann, 2011b): starting from a characteristics of developing countries in these respects high base that reflects the high cost of short-term credit, are the greater intensity with which problems manifest they are short (because the market is incomplete in themselves and the extent of the constraints they place long-term segments) and steep (because of the chronic on development. reluctance of the financial system to engage in longer-term On the Keynesian view, as has been shown, the lending, which is also manifested in the high cost of the financial system tends to operate in a way that is largely loans that are made despite this rationing) (see figure 1). dysfunctional to economic development in any country. It is widely recognized, for example, that capital markets FIGURE 1 are difficult to develop anywhere because of the greater Developed and developing countries: risks and costs they involve relative to the banking sector financing conditions (Herring and Chatusripitak, 2000). Financing difficulties and even financial exclusion in higher-risk segments are Developing countries also found in industrialized countries (Dymski, 2005 and Rates 2006). In developing countries, however, lower incomes Countries with developed and economic growth difficulties constrain the expansion nancial systems of the financial system in lower-risk segments too, so that even short-term credit is scarce and dear. Besides, higher-risk groups are a larger share of the population in less-developed countries, which aggravates the resource scarcity affecting them. Maturities Under these conditions, yield curves in developing Source: prepared by the author. countries acquire a profile that is quite unfavourable to

As for the external constraint, given the growing 3 Other conditions not directly related to development financing are importance of international trade and financial often identified as causes or manifestations of the relative backwardness relationships, any country’s development process is of developing countries. These conditions include: (i) the large share of gdp represented by goods with low value added, reflecting lesser shaped (and ultimately constrained) by the performance access to more elaborate technologies; (ii) poor production and urban of its balance of payments. In developing countries, infrastructure and inadequate education and health services, which though, this constraint tends to take the form of a chronic feed back into technological underdevelopment; (iii) low wages, reflecting a poorly trained workforce and a low investment rate; problem because it derives from a structural situation, (iv) a greater tendency to inflation in both growth and recession namely greater dependence on external financing, either stages, driven by a number of factors that include the high propensity to make up for the incompleteness of the financial system to consume typical of low-income economies; (v) low productivity; (vi) distributional conflicts, and (vii) frequent currency devaluations or to cover balance-of-payments deficits associated with to cope with the external constraints. technological underdevelopment. IV The functionality of the financial system to economic development

A basic condition for a financial system to operate in equated with the concepts of efficiency and financial a manner functional to economic development in any development as these terms are usually employed. The country is the availability of a diversified set of financial former, which is common to neoclassical approaches, institutions and instruments capable of meeting both the emphasizes the microeconomic functions of the financial varied demands associated with this process and the system, the aim of which is to meet the profit objectives profit objectives of the financial institutions themselves of financial institutions and enhance their clients’ (Carvalho, 2010). This condition, however, cannot be wealth. The efficient market hypothesis maintains that

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 74 CEPAL REVIEW 114 • DECEMBER 2014 performing these functions efficiently is a necessary “Functionality is defined as follows: a financial and sufficient condition for the financial system to also system is functional to the process of economic perform its macroeconomic function of providing finance development when it expands the use of existing for production activity. resources in the process of economic development The Keynesian approach does not deny the with the minimum possible increase in financial importance of these microeconomic functions, but it fragility and other imbalances, that may halt the rejects the efficient market hypothesis and, with it, the process of growth for purely financial reasons” equivalence between microeconomic and macroeconomic (Studart, 1995, p. 64). efficiency in the financial system. Besides the natural The special conditions of developing countries tendency of financial systems towards incompleteness suggest two extensions of Studart’s concept. First, when it comes to meeting economic development needs, the expansion of existing resources needs to meet there is the fact that the conditions determining the certain allocation criteria to reduce credit rationing efficiency of the financial system are not all necessarily and the financial exclusion affecting sectors essential met at the same time. A financial system can perform to development. A financial system that is functional the financing function with a structure that is both to economic development in developing countries undiversified (with a heavy concentration of commercial needs to have a variety of financing mechanisms to banks, for example) and inefficient at assessing risk. This support: (i) investment in innovations in all their phases; was the contradiction identified in developing countries (ii) expansion and modernization of productive and by the theorists of financial liberalization (Shaw, 1973; urban infrastructure; (iii) small and medium-sized McKinnon, 1973), who attributed the problem to the enterprises (smes); (iv) sectors with the characteristics policy of financial repression, i.e. to strong State action of a public good; (v) home ownership, and (vi) durable through regulation and public credit programmes. goods consumption. The low-income population should It is also possible for a financial system to meet the be included in the last two cases. objective of increasing wealth without properly performing Where the financing of innovations is concerned, the financing function. This has been the case in many given the high degree of uncertainty and credit rationing developing countries since the liberalization policies of involved, the experience of more developed countries the 1990s, which sought to expand and diversify financial points to the capital market as the best channel, operating systems but left many existing financing difficulties essentially through venture capital funds. Thus, meeting in place, including financial exclusion and external the requirement of functionality involves a more specific dependency (Hermann, 2004). requirement for the structure of the financial system: The functionality of the financial system cannot be capital market development. equated with the idea of financial development either, The second needful extension of Studart’s concept when the term is understood, as it usually is, in a purely concerns financial fragility: to be functional to development quantitative sense. On this definition, development is in developing countries, a financial system needs to have measured by aggregated indicators such as the number mechanisms to monitor external financial fragility in the and types of financial institutions, the value added area of what the author calls “other imbalances”. By of their assets, the size of the sector and the scale of analogy with Minsky’s original concept, this fragility is lending as a share of gdp, etc.4 Although important, these determined by currency mismatches, which can result indicators are not enough to capture the broader concept from the build-up of foreign currency borrowings by of macroeconomic functionality. According to Keynes’s agents whose income is in local currency (i.e. everyone and Minsky’s approaches, a financial system functional except exporters) or from maturity mismatches in to economic development needs to be able to fulfil two exporters’ external debts. essential conditions in addition to the physical existence From the Keynesian theoretical perspective, of a diversified financial structure: (i) the ability to meet therefore, the concept of functionality needs to the demand for financial resources in its various forms, encompass four aspects: (i) the volume of resources; encompassing both finance and funding needs, and (ii) (ii) allocation to sectors that are strategically important the ability to control the level of financial fragility, given to development; (iii) domestic financial fragility, and that some fragility is inevitable: (iv) external financial fragility. Although these dimensions may be dealt with separately for analytical purposes and usually require different instruments for good results, 4 An example of this type of approach can be found in idb (2005, they are not independent of one another, as the absence chap. 1).

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 75 of one may compromise the rest. For example, excessive Except where there are barriers (economic or otherwise) credit expansion and misallocation of resources (with to foreign capital, these phases result in a large expansion maturities shorter than the ideal or very high costs) of external borrowing in developing countries. Thus, some increase financial fragility in the economy. Fragility can of the financial fragility of these countries is manifested spread to the external sector if these financial system as external fragility. failings are compensated for by high external borrowing. This interpretation is not meant to suggest that the Lastly, a high degree of financial fragility or external volume of resources is irrelevant but only to emphasize fragility in the economy tends to restrict the supply of that, taken alone, it says little about the functionality funds, making it hard for the resource volume condition of financial systems in developing countries. Resource to be met. volume growth is supposed to be an indicator of improving The issues of allocation and external fragility are financing conditions in any country, but this hypothesis especially important in developing countries. Given is only borne out if the expansion is properly directed, the chronic difficulty of raising funds in the domestic reducing the degree of credit rationing and financial market, external fragility becomes an almost inevitable exclusion without unduly increasing financial and tendency in phases of abundant international liquidity. external fragility.

V Policies to enhance the functionality of financial systems in developing countries

A major practical implication of Keynesian financial short- and long-term private financing sources, summed theory is that, given the natural tendency of the financial up in high, short and steep yield curves. system towards incompleteness and financial fragility, The incentive policies needed to improve the it is unlikely to qualify as functional to economic situation would have to include measures to stimulate development in the absence of State action to achieve demand for financial assets, particularly long-term ones, this (Carvalho, 2010; Hermann, 2011a) in the form including variable-income securities (shares), in the of: (i) policies to encourage the development of the capital market. As regards compensatory policies, what financial system with a view to making it less incomplete; are recommended are directed private-sector lending (ii) policies to control domestic and external financial programmes and, most particularly, public credit policies fragility, always recognizing that these tendencies cannot targeted on sectors essential to development in each be wholly eliminated from the financial system, and period, with pricing and maturities that are less onerous (iii) compensatory policies to supplement the other two than the terms available in the market (in segments where kinds. This section discusses possible lines of action this option exists). in each of these areas, taking the four dimensions of As will now be argued, such policies, taken together, financial functionality for developing countries referred have the effect of lowering, extending and flattening to above. yield curves in the different segments of the financial system. In addition, they indirectly serve to reduce the 1. Policies on the amount and financial fragility involved in economic development, allocation of resources as better maturity and cost conditions reduce credit risk. Demand for a given asset is shaped by estimates of The policies needed to induce the financial system its likely risks and returns relative to alternative assets; to expand its operations in segments essential to by the macroeconomic environment, there being an economic development have to be guided by the extent increased preference for liquidity in situations of great and characteristics of financial system incompleteness uncertainty and for longer-dated securities at times of in each developing country. Despite these countries’ greater optimism; and by the incentive structure that specific characteristics, their financing difficulties have financial policy creates (Carvalho, 2010). Besides a features in common: the scarcity and costliness of both macroeconomic environment of low uncertainty, what

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 76 CEPAL REVIEW 114 • DECEMBER 2014 is needed to stimulate demand for long-term assets is assets, including longer-dated ones. The constraints on this an incentive structure that produces some comparative policy are, naturally, the risks of inflation, overborrowing advantage for these assets over lower-risk ones. This (and thence financial crises) and balance-of-payments structure can be created in two ways, which are not disequilibria. Nonetheless, the perverse effects of a high mutually exclusive: increased risk-adjusted rates of interest-rate policy are such that, even given these risks, return for long-dated securities, and lower rates for it should be used with diffidence and on a temporary short-dated securities. The profile of the yield curve basis and be supplemented by other instruments that in developing countries points to the latter as the most act more directly on these macroeconomic disequilibria advisable course. (discussion of which is beyond the scope of this article). Reducing short-term interest rates favours longer- In summary, this paper rejects the idea, upheld by dated operations (including the capital market) in two the Shaw-McKinnon model and widely propounded ways: (i) it will lower the yield curve, thus reducing the in this debate, that high interest rates might contribute costs of all financial operations, including long-term to economic development, whether in developing or ones, by also causing the curve to lengthen, and (ii) developed countries, by expanding the supply of loanable it will flatten the yield curve, reinforcing the previous funds. According to the Keynesian approach, higher effect, since lower returns on short-dated securities will interest rates can, at least in theory, increase the profits tend to increase investors’ preference for longer-dated of financial institutions, thereby stimulating lending. assets, which will provide the only route to increased This possibility, however, is subject to two types of risk: profitability for their portfolios in the new environment a rapid rise in defaults by borrowers and the consequent of low interest rates (figure 2). risk of a financial crisis. In both cases, the profitability of financial institutions will be negatively affected and credit growth will be cut short. Thus, it is considered that FIGURE 2 a policy of low short-term interest rates would do more Developing countries: the effects of lower short-term interest rates on the yield curve good than harm to the economic development process: in conjunction with the risk spreads of each form of Original curve financial operation (reduced, ideally, by macroeconomic policy), such interest rates would result in final rates Rates Lowering effect that were reasonable for borrowers and compatible with the levels of returns that financial institutions consider Flattening effect reasonable and satisfactory. Lastly, what a directed credit policy requires first and foremost is the creation of secure long-term funding sources to ensure that the private- or public-sector banks

Maturities discharging this function do not end up creating the very Source: prepared by the author. financial fragility (and fiscal fragility too in the case of public-sector banks) that they ought to be mitigating. Given the environment of constrained demand for long-term A systematic policy of encouraging low short- assets in which these policies have to be implemented, term interest rates also contributes to a macroeconomic the best source of long-term funding to underpin them environment that is more favourable to financial are non-market resources, i.e. funds that are fiscal or development. High interest rates tend to have a perverse parafiscal (compulsory saving) in origin. effect on the demand for assets in general because of Second, in practice these policies can only succeed adverse selection, which increases credit risk for lenders in compensating for the lack of lending to sectors deemed (Stiglitz and Weiss, 1981) and potentially compromises strategically important to development if credit is actually returns. Uncertainty and the preference for liquidity provided when required. In this respect, public-sector increase in this context, penalizing longer-dated assets credit tends to be more effective than directed credit. in particular. The latter, although it may help to make the market less Short-term rates that are appropriately low (i.e. that incomplete, suffers from a limitation common to more do not create the macroeconomic imbalances mentioned indirect forms of State action in the financial system below) tend to reduce borrowing risks and, by stimulating (supervision, regulation and incentives), in that it is a economic growth, improve expectations of returns on non-coercive policy which only acts through incentives.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 77

Directed credit programmes create rules for allocation savers and, at least in part, for any resultant losses. from particular private-sector funding sources, or make Besides protecting savers, which is necessary in itself, public fiscal or parafiscal funds available at below-market such regulation contributes indirectly to the development costs so that these institutions can provide the loans in of the country’s financial system by making the market question. These measures are meant to induce private- more attractive to savers in general. sector institutions to provide credit, but do not force them As for financial fragility, although prudential to, so their effectiveness may be limited by the same lack regulation can only deal with specific operations, its final of interest that caused the market to be incomplete in objective from the point of view of the functionality of the first place. Thus, the only sure way to improve the the financial system must be to control systemic risk, situation is for the government to directly take on the i.e. the degree to which the economy is exposed to risk in risks the private sector rejects. This requires the creation the financial system. As well as the risks taken or caused of development banks, i.e. public-sector banks with by specific agents, the possible channels of contagion specific functions in the economic development process. between them are also important. Controlling systemic risk requires restrictions on the build-up of speculative 2. Policies on financial fragility debts (including Ponzi schemes), as a way of managing risk-taking, and on the leverage of financial institutions, (a) Domestic financial fragility to limit contagion between them. Financial fragility can be partially controlled by Agents’ borrowing profiles and the channels market hedging mechanisms such as highly liquid assets, of contagion in the financial system are not clearly organized secondary markets, market makers, futures perceptible in the economy, however, which means that markets and derivatives. For these operations to take despite the macroeconomic cost involved in restricting place, however, it is necessary for the financial system the potential growth of the whole market, action to to have a reasonably well-developed capital market, forestall systemic risk needs to rely essentially on which is rarely the case in developing countries. This control of the overall risk exposure of the financial limitation reinforces the need referred to earlier for these system. Historically, such control has involved subjecting countries to coordinate allocation policies with those financial institutions to regulatory barriers and costs designed to control financial fragility. associated with the expansion of their lending operations Even in countries that meet this condition to some (Carvalho, 2005). degree, however, the fact that hedge operations, like the Regulatory barriers usually take the form of market ones covered by them, are based on expectations means segmentation preventing commercial banks from operating that these market instruments may prove inadequate or even in the capital market, along the lines of the Glass-Steagall counterproductive as a financial system defence measure. Act of 1935 in the United States. The main objective This is so, for example, in situations of particular asset of such a ban is to prevent contagion between the two price volatility, which encourage speculative operations segments of the financial system. Although successful in aimed rather at profit than at protection, and those where the United States and several other countries up until the herd behaviour takes over, since volatility makes it hard 1980s, this model fell into disuse from the 1990s with to arrive at reliable estimates (Kregel, 1997). the spread of liberalization policies (likewise initiated in Under such conditions, what is required if these the United States), which promoted the gradual removal instruments are to be negotiated with reasonable security of regulatory barriers to the free choice of portfolios by is not just physical infrastructure in the form of financial financial institutions. institutions and markets but also an institutional (regulatory) In this new financial system model, the only option infrastructure operating on two fronts: (i) preventive left is to impose costs on the choices with the greatest control of risk-taking in the financial system, and potential to generate systemic risk, essentially those (ii) compensation of the macroeconomic effects when involving the greatest liquidity risk (Carvalho, 2009). preventive measures prove inadequate or ineffective. These costs have taken a variety of forms (Carvalho, Preventive regulation needs to focus on protection 2005): the requirement for minimum levels of liquidity for the end saver and control of financial fragility. What and capital on the balance sheets of financial institutions is basically required for the former is a regulatory (compulsory levies on deposits, loan provisions, Basel apparatus that establishes the responsibility of financial rule, etc.), similar regulation of the derivatives market institutions for providing information on the risks of (off-balance sheet operations) using liquidity and capital operations they undertake using resources raised from buffers, and tax measures, among others.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 78 CEPAL REVIEW 114 • DECEMBER 2014

In developing countries classified as emerging “Innovation (both financial and organizational) […] markets, where capital markets are reasonably well is in itself a source of institutional unsuitability. […] developed, prudential regulation needs to be extended [U]nder the effect of the development of financial to both the corporate debt and the share segments of systems, formal institutions […] become ineffective these markets. Corporate debt is similar to bank credit, and no longer sufficiently counter the endogenous except for the greater liquidity of corporate securities, dynamic of risk taking” (Sinapi, 2011, p. 18). even though this is typically not very high either (Herring These limitations suggest the advisability of applying and Chatusripitak, 2000). Shares do not increase the level compensatory policies of more direct intervention in the of debt in the economy, and indeed are an alternative financial system to supplement prudential regulation, to debt, but nor are they free from systemic risk. This such as: (i) regulatory barriers to risk-taking by financial essentially derives from: (i) the market risk typical institutions; (ii) emergency hedging instruments for of shares; (ii) this segment’s greater exposure to herd times of particular uncertainty (emergency open market behaviour, facilitated by the liquidity of shares relative operations using public securities and currency, for to other types of assets, and (iii) the fact that many example); (iii) lending of last resort by the central investors, and especially larger ones, are leveraged, bank to financial institutions with temporary liquidity i.e. work with borrowed resources, creating the risk problems; (iv) programmes to restructure the banking of banking-sector contagion in the event that the share sector when there are prolonged difficulties indicative of market performs poorly. solvency problems, and (v) circuit breaker mechanisms Controlling the systemic risk associated with the in the capital market implemented by official regulators capital market thus requires instruments similar to those or by private-sector self-regulation bodies. applied to banks: requiring intermediaries to disclose information accurately and making them liable for any (b) External financial fragility losses that result from failures in this regard; requiring As has been pointed out, the external financial liquidity and capital buffers; and limiting financial fragility of developing countries is structurally determined institutions’ leverage and holdings of securities on their by the profile of their production structure and the balance sheets. particularly fragile position they thus occupy in the The effectiveness of the cost-based risk control international financial market. These conditions are model is limited, however, by factors that are largely beyond the scope of financial policy action, which beyond the control of the regulatory authorities, such does however play an important role in controlling the as the phase of the business cycle and ongoing financial systemic risk associated with possible contagion in the innovation. Where the business cycle is concerned, domestic financial system when a country’s external prudential regulation can be expected to be least fragility results in liquidity difficulties for financial effective in expansion phases when markets are most institutions. This contagion can take place via three bullish, which is just when financial expansion most channels: the external liabilities of financial institutions needs controlling. The financial system tends to play themselves, those of the non-financial sector (including down the burden of regulatory costs in this situation, balance sheet and off-balance sheet operations in both expecting them to be more easily offset by the greater cases), and the currency market. earnings forecast. This limitation can be eased, although The first of these produces contagion, in particular, not wholly overcome, by a countercyclical structure from the financial system to the balance of payments, since for regulatory costs based, for example, on a sliding in this case the liabilities of the financial system are one scale of costs that rises with the volume of resources or of the causes of the country’s external fragility. Contagion with the rate of expansion of each type of operation to through the second channel arises as a counterpart of be controlled. the external fragility of the non-financial sector, whose Financial innovations place a twofold limitation demand for currency can unleash a large-scale movement on the effectiveness of controls designed to prevent to redeem domestic financial investments, increasing financial fragility by imposing costs. First, they gradually banks’ very short-term liabilities (in reserves). Systemic make the current cost structure obsolete, as certain crises arising through these two channels can be avoided types of operation become less of a feature of the if the flow of external liabilities is compatible with the market. Second, they make it hard to recalibrate control volume of currency available in the country. Thus, the instruments, since the new types of operation are still systemic risk associated with these channels can be relatively unknown: contained by a strategy similar to the one identified earlier

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 79 for controlling domestic financial fragility: regulatory in developing countries in the 1990s triggered a series limits on the accumulation of external liabilities by the of currency crises there and led to the old model being financial system and the non-financial sector, conjoined replaced with floating currency regimes. with differentiated control measures that should focus Exchange-rate volatility is a fact of life in floating on short-term transactions, as these keep demand for currency regimes since, by definition, the central bank currency systematically high. does not adopt any formal commitment on the exchange Contagion through the currency market takes rate. On the other hand, an unquestionable commitment place when adverse exchange-rate movements are able by any government to macroeconomic stability, which to inflict major losses on the financial system. Given implies some degree of monetary and financial-system that developing countries are more likely to suffer stability, requires central banks operating hybrid from external fragility, exchange-rate depreciation is regimes with flexible exchange rates to engage in the particularly risky because it raises the cost of a country’s task of containing currency volatility. This gave rise to external liabilities (including the financial system’s) the dirty float regimes that became predominant in the and aggravates the effects from the other two channels. late 1990s. Currency appreciation, however, is not necessarily In this model, exchange-rate volatility can be beneficial for a developing country’s financial system, controlled by central bank interventions in spot and as it will tend to worsen its international trade position, futures markets involving the trading of currencies and thereby exacerbating external fragility. Moreover, in currency derivatives. In spot operations, the bank acts developing countries with reasonably developed futures directly on the current exchange rate, while in the futures markets, a category that includes most emerging markets, market it influences exchange-rate expectations. In neither both currency depreciation and appreciation can inflict case, however, is the central bank wholly immune to large losses on the financial system in the event that it difficulties deriving from the great mobility of capital, is heavily committed to currency derivatives, since it is which limit its ability to control the exchange rate in enough for the direction or strength of the exchange- managed currency regimes. Specifically, in situations of rate movement to take a large part of the market great uncertainty, herd behaviour in the financial system by surprise. limits the central bank’s ability to act in the face of strong In this context, there needs to be a permanent pressure for currency appreciation or depreciation. Given policy of controlling exchange-rate volatility in addition that ultimately these situations cannot be avoided, it is to the limits on external liabilities already referred to. advisable for central banks not only to act on the price The latter reduce the exposure of the financial system of the currency, but also to introduce mechanisms to to exchange-rate risk, while controls on volatility, by control the capital flows influencing this. making the exchange rate more predictable, tend to Selective countercyclical control of capital inflows reduce the currency risk entailed in any kind of operation is particularly important in developing countries, given that creates external liabilities. their greater propensity to borrow abroad at times when In the managed exchange-rate regimes that prevailed foreign investors are particularly bullish. Regulatory in developing countries until the late 1990s, currency or tax barriers to capital inflows at these times, with volatility was avoided by the constant action of central greater restrictions on short-term operations, help to banks committed to keeping the exchange rate at the level forestall excessive expansion of external borrowing. In set. As second-generation models indicate (Obstfeld, very bearish phases, likewise, selective controls can be 1994), the sustainability of these regimes depends on applied to discipline the movement of capital outflows the confidence of the market, and financial institutions from the country. In both cases, the country’s external in particular, in the ability of the central bank to meet fragility and the volatility of the currency market the exchange-rate target announced. If this condition is are mitigated. not met, the country becomes vulnerable to speculative Note should be taken, finally, of the risk that attacks (such as a sudden increase in the preference for policies to control financial fragility may conflict with foreign currency over local-currency investments) strong development financing needs, as they tend to restrict the enough to force the central bank to alter the exchange- total volume of operations in the financial system. This rate target or, in extreme cases, to abandon the managed dilemma shows the extreme importance of coordinating exchange-rate regime altogether. As is well known, the such control with the allocation policies mentioned severe shock to the credibility of these regimes from the earlier in order to shape a financial system that is more trade and financial liberalization policies implemented functional to development.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 80 CEPAL REVIEW 114 • DECEMBER 2014

VI Conclusions

To be functional to economic development, a financial Accordingly, there is also a need for compensatory system must be capable of meeting its financing needs policies of more direct intervention wherever stimulation at acceptable cost and with appropriate maturities. The or restraint is called for. This group includes, in particular: absence of these conditions does not necessarily prevent (i) programmes of directed public- or private-sector the process from moving forward, but it does give it a credit in the field of allocation policies (with beneficial “stop and go” character, as the financing structure will indirect effects for the control of financial fragility as suffer from a high degree of financial fragility due to the well); (ii) non-market regulatory barriers in the field of preponderance of high-cost short-term debt, and perhaps financial fragility control policies, and (iii) countercyclical external fragility too, in cases where the shortcomings measures to restrict the free flow of capital in order to of the domestic financial system are compensated for control external fragility. by external borrowing. Thus, a functional financial Interventionist policies are not risk-free either. The system is not just a desirable condition that facilitates difficulties encountered by agents generally in predicting economic development but a necessary one if this process the behaviour of relevant indicators also affect the is not to turn into a source of systemic risk and, in all government bodies responsible for implementing these probability, of banking and currency crises that impose policies. Even so, the financial history of the twentieth major sacrifices on the economy. century and the early decades of this one suggests that the From a Keynesian perspective, as has been shown, macroeconomic damage deriving from this forecasting the normal operating conditions of the financial system risk could be less than that deriving from market risk.5 tend to make it generally incomplete and dysfunctional, During the five decades the interventionist model even in countries at an advanced stage of economic and lasted (from the mid-1930s to the 1980s), no systemic financial development. Thus, it is unlikely that these crisis struck the world economy, despite the indisputable conditions will be achieved without State policies designed progress of economic and financial development. In that for the purpose. In developing countries, the difficulties period, the world experienced two mortgage crises (during are compounded by the very conditions that mark them the 1970s and 1980s) originating in rich countries with out as less developed, in particular a lack of structural great influence on the international market, the United diversification in the financial system and intermittent States and Japan, plus the Latin American external debt dependence on external financing. These conditions make crisis of the 1980s. Although these were very painful developing countries more subject to periods of growth for the countries affected, the regulatory barriers of without development (Furtado, 1967) and to financial the time meant that none turned into an international fragility, especially in its external dimension, creating a systemic crisis. vicious circle that perpetuates their unfavourable position The events of recent years, however, have not been a on the international stage. good advertisement for the incentive-based conventional Improving the situation requires a coordinated set financial policy model that has predominated in the era of of financial policies, resting on two pillars: (i) allocation liberalization. In a little over 20 years of existence, this policies, focusing on segments of strategic importance model has not been able to prevent two systemic crises to economic and financial development, and (ii) policies of international reach, coming against a background to control financial fragility, both domestic and external. of lower economic growth than in the earlier period: Considering, however, that even large and diversified the crisis in developing Asia, which began in 1997 and financial systems still show signs of behaviour that is spread to Latin America between 1999 and 2003, and dysfunctional to economic development (such as a lack of interest in projects with the character of public goods, financial exclusion, etc.), incentive- and cost-based 5 The exception is the political risk (of corruption, for example) conventional policies aimed at deregulated markets associated with any kind of government action. This is not a specific problem of financial policy, however, or even of economic policy in will not be enough to induce an adequate degree of general, but is of a political and legal nature and ought to be dealt functionality in developing-country financial systems. with in that context.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN CEPAL REVIEW 114 • DECEMBER 2014 81 the crisis that began in 2008 (symptomatically enough to associate these crises with the shift in the model for in the “cradle of liberalization”, the United States) and controlling systemic risk that resulted from financial is still ongoing as of 2013. The latter crisis reinforces liberalization. Recent events make it advisable, at the the hypothesis that the current model is ineffective, as very least, for the subject of interventionist policies it has been severest in rich countries with developed (which need not necessarily be in the same mould as the financial systems. old ones) to be put back on the agenda of the continuing Without ignoring the differences in macroeconomic debate about possible ways of making the financial system context that make linear comparison between the periods more functional to economic development, not just in and countries concerned impossible, it is difficult not developing countries, but in industrialized ones too.

Bibliography Cargill, T. (1983), Money, the Financial System and Monetary Policy, (2004), “O trade off do crescimento no Brasil nos anos Englewood Cliffs, New Jersey, Prentice-Hall. 1990-2000: análise crítica e alternativas de política monetária”, Carvalho, F.J. Cardim de (2010), “Reformas financeiras para apoiar Revista Econômica, vol. 6, No. 2, Rio de Janeiro, December. o desenvotlvimento”, Textos para Discussão cepal-ipea, (2003), “O modelo de liberalização financeira dos anos No. 14 (LC/BRS/R.231), Brasilia, Economic Commission for 1990: ‘restatement’ ou auto-crítica?”, Nova Economia, vol. 13, Latin America and the Caribbean (eclac)/Institute of Applied No. 2, Belo Horizonte, Federal University of Minas Gerais, Economic Research (ipea). July-December. (2009), “Systemic crisis, systemic risk and the financial Herring, R.J. and N. Chatusripitak (2000), The Case of the Missing instability hypothesis”, Macroeconomic Policies on Shaky Market: The Bond Market and Why It Matters for Financial Foundations, E. Hein, T. Niechoj and E. Stockhammer (eds.), Development, Tokyo, Wharton Financial Institutions Center, May. Berlin, Metropolis-Verlag. idb (Inter-American Development Bank) (2005), Relatório (2005), “Inovação financeira e regulação prudencial: da 2005. Libertar o crédito: como aprofundar e estabilizar o regulação de liquidez aos Acordos de Basiléia”, Regulação financiamento bancário, Rio de Janeiro, Campus. financeira e bancária, R. Sobreira (org.), São Paulo, Keynes, J.M. (1943), General Theory of Employment, Interest, and Editora Atlas. Money, London, Macmillan. (1997), “Financial innovation and the post Keynesian (1937a), “Alternative theories of the rate of interest”, The approach to the ‘process of capital formation’”, Journal of Post Economic Journal, Hoboken, New Jersey, Wiley, June. Keynesian Economics, vol. 19, No. 3, M.E. Sharpe. (1937b), “The ‘ex-ante theory’ of the rate of interest”, The Dymski, G.A. (2006), “Discrimination in the credit and housing Economic Journal, Hoboken, New Jersey, Wiley, December. markets: findings and challenges”, Handbook on the Economics Kregel, J. (1997), “Margins of safety and weight of the argument of Discrimination, William Rodgers (ed.), Cheltenham, United in generating financial fragility”, Journal of Economic Issues, Kingdom, Edward Elgar. vol. 31, No. 2, Association for Evolutionary Economics, June. (2005), “A eficiência social e a regulação bancária: lições (1986), “A note on finance, liquidity, saving, and da experiência americana”, Regulação financeira e bancária, investment”, Journal of Post Keynesian Economics, vol. 9, R. Sobreira (org.), São Paulo, Editora Atlas. No. 1, Armonk, M.E. Sharpe. Fry, M.J. (1995), Money, Interest and Banking in Economic (1980), “Markets and institutions as features of a capitalistic Development, London, Johns Hopkins University Press. production system”, Journal of Post Keynesian Economics, vol. 3, Furtado, C. (1967), Teoria e política do desenvolvimento econômico, No. 1, Armonk, M.E. Sharpe. São Paulo, Companhia Editora Nacional. Malkiel, B.G. (1994), “Efficient market hypothesis”, The New Gerschenkron, A. (1962), Economic Backwardness in Historical Palgrave Dictionary of Money and Finance, P. Newman, M. Perspective: A Book of Essays, Cambridge, Massachusetts, Milgate and J. Eatwell (eds.), London, Macmillan Press Limited. Belknap Press of Harvard University Press. Matouk, J. (2010), “Financing innovation: an historical approach”, Gurley, J. and E. Shaw (1960), Money in a Theory of Finance, Journal of Innovation Economics, vol. 6, No. 2, De Washington, D.C., Brookings Institution. Boeck University. (1955), “Financial aspects of economic development”, McKinnon, R. (1973), Money and Capital in Economic Development, The American Economic Review, vol. 45, No. 4, Nashville, Washington, D.C., The Brookings Institution. Tennessee, American Economic Association, September. Minsky, P.H. (1986), Stabilizing an Unstable Economy, New Haven, Hall, B.H. (2002), “The financing of research and development”, Yale University Press. Working Papers, No. E01-311, Berkeley, University of _____ (1982), “The financial fragility hypothesis: an interpretation of California, Berkeley, January. Keynes and an alternative to ‘standard’ theory”, Can It Happen Hermann, J. (2011a), “Política financeira e desenvolvimento Again? Essays on Stability and Finance, H.P. Minsky, New econômico: um debate a partir de contribuições recentes York, M.E. Sharpe, Inc. de Fernando Cardim de Carvalho”, Anais do IV Encontro Obstfeld, M. (1994), “The logic of currency crises”, nber Working Internacional da Associação Keynesiana Brasileira, Rio de Paper, No. 4640, Cambridge, Massachusetts, National Bureau Janeiro, Institute of Economics, Federal University of Rio de of Economic Research. Janeiro, August. Schumpeter, J.A. (1934), The Theory of Economic Development, (2011b), “O mercado de títulos privados e a inadequação da Cambridge, Massachusetts, Harvard University Press. curva de rendimentos no Brasil: análise teórica e da experiência Shaw, E.S. (1973), Financial Deepening in Economic Development, pós-estabilização (1995-2010)”, Working Paper, No. 14, New York, Oxford University Press. Brazilian Development Bank (bndes)/National Association of Sinapi, C. (2011), “Institutional prerequisites of financial within Postgraduate Centers in Economics (anpec). Minsky’s financial instability hypothesis: a proposal in terms

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN 82 CEPAL REVIEW 114 • DECEMBER 2014

of ‘institutional fragility’”, Working Paper, No. 674, Dijon, (1995), Investment Finance in Economic Development, Levy Economics Institute of Bard College, July. London, Routledge. Stiglitz, J. (1994), “The role of the state in financial markets”, (1995-96), “The efficiency of financial system, liberalisation Proceedings of the World Bank Annual Conference on and economic development”, Journal of Post Keynesian Development Economics 1993, Washington, D.C., World Bank. Economics, vol. 18, No. 2, M.E. Sharpe Inc. Stiglitz, J. and A. Weiss (1981), “Credit rationing in markets Tavares, M. da C. (1979), Da substituição de importações ao with imperfect information”, American Economic Review, capitalismo financeiro, Rio de Janeiro, Zahar Editores. vol. 71, No. 3, Nashville, Tennessee, American Economic Tobin, J. (1958), “Liquidity preference as behavior toward risk”, The Association, June. Review of Economic Studies, vol. 25, No. 2, Oxford University Studart, R. (2005), “Financiamento do desenvolvimento”, Economia Press, February. Brasileira Contemporânea: 1945-2004, F. Giambiagi and Zysman, J. (1983), Governments, Markets and Growth, London, others (orgs.), Rio de Janeiro, Campus. Cornell University Press.

FINANCIAL CONSTRAINTS ON ECONOMIC DEVELOPMENT: THEORY AND POLICY FOR DEVELOPING COUNTRIES • JENNIFER HERMANN The impact of China’s incursion into the North American Free Trade Agreement (nafta) on intra-industry trade

Jorge Alberto López A., Óscar Rodil M. and Saúl Valdez G.

ABSTRACT China has become a major player in world trade. Although it has not signed any trade agreements with the countries of the North American Free Trade Agreement (NAFTA), China has been gaining ground as a supplier of goods, making vigorous inroads into this area. One of the dominant trends in economic integration has been the development of

intra-industry trade, which has flourished in thenafta signatory countries. This paper focuses on the analysis of intra-industry trade in the context of this free trade area, where the production structure of the countries involved has changed significantly since trade liberalization, revealing the internationalization of production chains. Lastly, changes in

the trade structure induced by the growing presence of China in the nafta region are captured. Trade within this area works like a radiated wheel, with the United States acting as the axis, while China, Canada and Mexico operate as the spokes.

KEYWORDS International trade, China, United States, Mexico, nafta, treaties, free trade, intra-industry trade, exports, imports, trade statistics

JEL CLASSIFICATION F14, F15, L16

AUTHORS Jorge Alberto López A. is a Professor-Researcher in Economics at the Faculty of Social Sciences of the Autonomous University of Chiapas, Mexico. [email protected] Óscar Rodil M. is a Professor-Doctor at the Faculty of Economic and Business Sciences of Santiago de Compostela University, Spain. [email protected] Saúl Valdez G. is a postgraduate student in economics at the Faculty of Economics of the National Autonomous University of Mexico, Mexico. [email protected] 84 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

This paper seeks to analyse the pattern of intra-industry to parts and components originating in China (Gazol, trade (exchange of varieties of the same good) in the 2007), which has gained market share in Mexico in context of the emergence of China in the area of the recent years.1 North American Free Trade Agreement (nafta). This As for the debate about the effects arising from country has become a prominent player in world trade, the integration process, different economic theories in particular in North America. Although this Asian abound and no definitive answer has been provided giant has not signed any trade agreements with the about the regional impact of the integration processes, nafta countries, it has entered into this area, gaining since the effects vary depending on the theoretical ground as a goods supplier. Our aim is to determine the approach adopted (Rodríguez-Pose and Petrakos, 2004). significance of these trends and their implications, with Therefore, empirical analysis is needed to establish the special attention to the intra-industry integration model. main economic changes linked to the nafta process, and In addition to analysing intra-industry trade flows in particular to demonstrate how China has integrated over the period 1993-2011, this study ties in with into the nafta zone as a goods supplier. the theoretical debate relating to the establishment The paper is structured as follows. Section II of economic areas. These integration processes have addresses the conceptualization and methodological sparked off heated theoretical controversy over the trade aspects of intra-industry trade. Section III examines effects they have generated. For a long time, pioneering China’s incursion into nafta and its repercussions authors such as Viner (1950) and Dornbusch (1992) have for complementarity and competition. Globalization, maintained that the member countries of the economic restructuring and models for the integration of two areas have achieved welfare gains at the expense of the emerging economies, in this case Mexico and China, are rest of the world. In a case such as nafta, reducing covered in section IV. Meanwhile, section V examines domestic barriers could boost competitiveness among China’s incursion into the North American market and member countries by increasing the relative efficiency of the ensuing changes to the nafta intra-industry trade the nafta economic area and intra-nafta imports and model. Lastly, the main conclusions of the study are exports. Hence, assuming the existence of economies of presented in section VI. scale, externalities and dynamic comparative advantages, intra-nafta imports would replace those from the rest of the world. However, empirical evidence suggests 1 Hence the successive revisions to Annex 401 of nafta in order to that the United States has lost market share in Mexico relax rules of origin; the geographical location of providers has been (Dussel and Gallagher, 2013), although this is due mainly changed but they remain the same providers (Gazol, 2007).

II Intra-industry trade: conceptualization and methodological aspects

Theoretically, the problem of intra-industry trade was international trade led to the economic concept now first discussed in the 1960s with the studies by Verdoorn known as intra-industry trade. (1960), Balassa (1963) and Grubel (1967). These authors Authors such as Krugman (1995) or Grossman found that a growing share of trade took place within and Helpman (1990) focused on intra-industry trade the same industries and sectors. This novel form of and discovered notable developments on the basis of

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 85 new theories of international trade. Modern trade theory intra-industry trade occurs when the same production provides a set of explanatory elements grounded in line is located in different countries, leading to the re- imperfect competition, economies of scale and different export of goods (Dussel and León González, 2001). varieties of goods. Intra-industry trade emerges as a From a methodological viewpoint, several indicators result of increasing returns, product differentiation and are designed to measure the degree of intra-industry consumers with diverse preferences. First, economies of trade in different economies, the most common being scale promote production concentration, meeting large the Grubel-Lloyd (gl) index.2 This index is built on the demands from a small number of production centres. basis of bilateral trade flows between countries. These Second, each company can differentiate its products flows can be divided into two groups: inter-industry from those of rival companies in order to segment trade (different goods) and intra-industry trade (similar demand and maintain a degree of monopoly over its goods). The gl index ranges from 0 to 1, depending own variety. Third, one prerequisite for intra-industry on the absence (gl = 0) or total occurrence (gl = 1) of trade is the existence of a consumer mass with different intra-industry trade.3 Frequently, an alternative adjusted preferences for the multiple product varieties offered. expression of the gl index is used at the aggregate level These three conditions tend to become more visible in to avoid the destabilizing effect of the trade balance.4 In the case of developed economies, which explains why addition, other authors (Cárdenas and Dussel, 2011) use some integration areas, such as the European Union, an index proposed by Hamilton and Kniest (1991) which have excelled in such exchanges. measures marginal intra-industry trade with respect to Several processes have developed in parallel with total aggregate trade. the rise of intra-industry trade and, in a sense, have been An important point to note is that the degree of driven by it. Progress in trade liberalization, particularly data disaggregation influences the measurement of of industrial products, has occurred both globally under intra-industry trade. It is best to use the largest possible the General Agreement on Tariffs and Trade (gatt) and sectoral disaggregation in order to accurately identify the World Trade Organization (wto) and regionally intra-industry trade flows. The statistics used in this study through the European Union, nafta, the Association provide a complete sectoral breakdown for the two-digit of Southeast Asian Nations (asean), the Southern level (tariff chapters) of the Harmonized Commodity Common Market (mercosur) and the Central American Description and Coding System, with information from Common Market (cacm). These processes have been the United Nations Commodity Trade Statistics Database guided primarily by the expansion of intra-firm trade in (comtrade). Information at the four-digit (heading) level pursuit of the free movement of intermediate and final is used only for the chapters where a higher intensity of goods to the benefit of multinational corporations. Some intra-industry trade is observed. authors (Navaretti, Haaland and Venables, 2002; oecd, It should be noted that this paper analyses 2002; Helpman, 2006) introduced this last aspect in the trade between countries with very different levels of analysis of intra-industry trade, focusing on the role development, for example, between the United States or of multinationals. Indeed, according to these authors, Canada, on the one hand, and China or Mexico, on the multinationals are the real protagonists of the current other. These combinations allow us to consider various globalization process. patterns of trade, including North-North, North-South To summarize, we can identify three models of and South-South. intra-industry trade: first, the model based on product differentiation and economies of scale (the most common); 2 second, the model of functionally homogeneous goods The formula of the aggregate index of Grubel and Lloyd is: gl = 1-[∑|xi-mi | /∑(xi+mi)], where xi and mi are the value of exports and (closely linked to border and periodic or seasonal trade); imports of sector i respectively. The adjusted version of the aggregate and lastly, the model based on the technology gap, the index of Grubel and Lloyd is expressed: gladjusted = [∑(xi+mi)-∑|xi- product life cycle and the internationalization of the mi|]/[∑(xi+mi)-|∑xi-∑mi|], where xi and mi are the value of exports and imports of sector i, respectively. production process (intra-firm). 3 The aggregate version of the gl index is: gl = 1-[∑ | xi-mi | /∑(xi+mi)], Another perspective is the differentiation between where xi and mi are the value of exports and imports of sector i, vertical and horizontal intra-industry trade. Horizontal respectively. intra-industry trade occurs when two separate production 4 The adjusted version of the aggregate version of the gl index is chains of the same industry and a similar level of expressed as gladjusted = [∑(xi+mi)-∑ | xi-mi | ]/[∑(xi+mi)- | ∑xi-∑mi | ], where xi and mi are the value of exports and imports of sector i, development exchange goods internationally. Vertical respectively.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 86 CEPAL REVIEW 114 • DECEMBER 2014

III China’s incursion into nafta: between complementarity and competition

Prior to the 1990s, China had shown little interest in and China has been growing rapidly, especially between signing formal agreements and pursuing regional trade; the United States and China, even though none of them this was partly because it was not yet ready to deal with has signed a free trade agreement with China. In fact, the rapid liberalization of trade and investment. With China has emerged as the leading supplier in the United the advent of the twenty-first century, however, China’s States market, replacing Mexico in 2003 and Canada attitude to regional cooperation changed dramatically in 2009 (see figure 1), and as the second supplier in the (Wang, 2004, cited by Yu, Xue and Hong, 2006). Joining Mexican market since 2002, ousting Canada (see figure 2). wto forced China to take on several commitments in The coverage ratio (defined as the percentage of terms of market liberalization, and integrated the country trade conducted under agreements in relation to total into the world economy (Yu, Xue and Hong, 2006). trade) stands at 11.2% for China, 34.4% for the United China is now a signatory to 10 free trade agreements States, 68.4% for Canada and 81.5% for Mexico (Rosales with 24 countries, including 3 from Latin America and Kuwayama, 2012) (see table 1). As these data show, (Chile, Costa Rica and Peru). Another three free trade China has had no need of free trade agreements to become agreements are under negotiation with Australia, Norway a world trade power. and Switzerland (wto, 2013). Analysts hold opposing views as to whether the From a theoretical point of view, one of the main role of China and Mexico in the United States market arguments in favour of free trade agreements is that they is competitive or complementary. Some authors, such seek to improve the economic dynamics of the signatory as Feenstra and Looi Kee (2009), estimate that there is countries. However, trade between the nafta countries increasing competition between these two countries for

FIGURE 1 United States: imports from China and its nafta partners as a share of total imports, 1993-2011 (Percentages)

25

20

15

10

5

0 199 3 199 4 199 9 200 0 200 1 200 6 200 7 200 8 199 5 199 6 199 7 199 8 200 2 200 3 200 4 200 5 200 9 201 0 201 1

China Canada Mexico

Source: United States Census Bureau, [online] http://www.census.gov/foreign-trade/statistics/country/, 2013.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 87

FIGURE 2 Mexico: imports from China and its nafta partners as a share of total imports, 1993-2011 (Percentages)

80

70

60

50

40

30

20

10

0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

China Canada United States

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade).

TABLE 1 naftaa countries and China: coverage ratio of trade agreements, 2009 (Percentages)

Free trade agreement coverage Country Exports and imports Exports Imports United States 34.4 40.1 30.5 Canada 68.4 77.7 59.2 Mexico 81.5 93.0 70.2 China 11.2 10.1 12.6

Source: Japan External Trade Organization (jetro), 2010 jetro Global Trade and Investment Report. A Global Strategy for Japanese Companies to Open New Frontiers in Overseas Markets, Tokyo, 2010; and O. Rosales and M. Kuwayama, China and Latin America and the Caribbean: Building a Strategic Economic and Trade Relationship, Libros de la cepal, No. 114 (LC/G.2519-P), Santiago, Chile, Economic Commission for Latin America and the Caribbean (eclac), March 2012. United Nations publication, Sales No. E.12.II.G.2. a North American Free Trade Agreement.

the United States market and that foreign direct investment do not compete with China in the United States market, (FDI) has been shifting towards China at the expense but that each country has its own niche, according to of Mexico (De la Cruz, Núñez and Ruiz-Porras, 2008, the specialization patterns that emerged in the 1990s. cited by De la Cruz and Marín, 2011). Feenstra and Looi Of particular interest in this paper is the analysis of Kee (2009) further argue that the difference between intra-industry trade, which should clarify to what extent Mexico and China is that the former slashed its tariffs these economies are complementary or competitive. This to an excessive degree while China did so unilaterally, point is raised with full knowledge of the fact that China protecting some branches such as agriculture. Neme has become a leading power globally, and especially in (2006), however, argues that Mexican manufacturers North America.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 88 CEPAL REVIEW 114 • DECEMBER 2014

IV Globalization and restructuring: models for the integration of two emerging economies, China and Mexico, into the world market

By the early 1980s, Mexico showed limited openness to Meanwhile, since 1978, China has implemented a trade. In fact, in 1983, virtually all imports into Mexico series of reforms that began with the four modernizations were subject to prior permission (non-tariff barriers) and advocated by Deng Xiaoping: agriculture, industry, extremely high tariffs. However, since that year, Mexico defence and science and technology (Neme, 2006). has restructured its economy in an attempt to achieve The open-economy policy led to the adoption of a three goals: sound public finances, privatization of State legal framework to facilitate international economic enterprises and trade liberalization. Rationalization of relations and foreign direct investment, the creation of trade protection transformed the country from one of special economic zones and open cities to modernize the most closed into one of the most open economies in domestic industry by establishing foreign companies the world. The opening has been such that in 2006 only that manufacture and export products helped by diverse 4.1% of the value of non-maquiladoras and 2.7% of total incentives (Neme, 2006, pp. 30-31). A clear sign of this imports were subject to prior permission. Moreover, the reform process is China’s increasing economic openness tariff average, which was 27% in 1982, fell to 5.9% in since 1993, expanding from 30% in that year to 77% in August 2012 (the weighted tariff in 1982 was 16.4%, 2011 (see figure 3). but in 2012, it was only 0.56%) (cefp, 2006; Gaceta In this context of opening and reform, China and Parlamentaria, 2012).5 A clear sign of this process is Mexico have become leading players in the global the increase in the openness ratio from 30% to 81% economy, making major advances in exports of between 1993 and 2011 (see figure 3). manufactures. China became a significant part of the In short, this illustrates the accelerated liberalization global factory when its share of world exports jumped of the Mexican economy, which is reflected in its from 2.8% in 1993 to 15.4% in 2011. Mexico’s share, increasing openness, tariff reductions and the dismantling meanwhile, increased from 0.60% in 1993 to 2.0% in of non-tariff barriers. In fact, after these changes, the 2011, and the country thus became the foremost Latin concentration of foreign trade of Mexico with the American exporter of manufactures and an important United States is two-thirds of total trade in Mexico at part of the “global factory”.6 present (although in some years it has accounted for In recent years, China has become a major player on more than 70%). This concentration deepened under the world stage and a regional economic power in Asia. the nafta process until the recession of 2001 and the Indeed, the country has undoubtedly become a first-order entry of China into wto in the same year, which had a global power. China’s share in world manufacturing dampening effect on this phenomenon. exports has surpassed that of all the nafta countries put together, making it a linchpin of the global factory (see figure 4).

5 Gazol (2007) discusses the return to a new phase of protectionism that permitted foreign purchases equivalent to 2.2% in 1995, compared with 10% and 11% in 2005 and 2006, respectively. However, this 6 Calculations based on the World Bank DataBank, in current United pattern differs from the one that prevailed in the 1980s. States dollars.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 89

FIGURE 3 Mexico and China: degree of economic openness, 1993-2011 (Percentages of gdp, on the basis of dollars at constant 2000 prices)

90

80

70

60

50

40

30

20

10

0 19 93 199 4 1995 19 96 1997 199 8 19 99 2000 20 01 200 2 200 3 20 04 2005 2006 200 7 200 8 200 9 20 10 2011

Mexico China

Source: prepared by the authors, on the basis of data from World DataBank.

FIGURE 4 China and naftaa signatory countries: manufacturing exports, 1993-2011 (Percentages of world total)

25

20

15

10

5

0 00 7 00 8 199 3 199 4 1995 199 6 199 7 199 8 199 9 200 0 2001 2002 2003 2004 2005 200 6 2 2 200 9 201 0 2011

China  Source: prepared by the authors, on the basis of data from World DataBank. a North American Free Trade Agreement.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 90 CEPAL REVIEW 114 • DECEMBER 2014

V China’s incursion into the North American market and ensuing changes to the nafta intra-industry trade model

From an empirical viewpoint, this paper seeks to analyse Intra-industry trade developed as part of an intensive the pattern of intra-industry trade during a period marked process of integration between Mexico, the United States by the integration process in North America (nafta) and Canada, at a time when China’s share in nafta and the emergence of China in this region. Previous imports was increasing. The high growth in trade and studies (Dussel and León González, 2001; oecd, 2002; the concentration of Mexican and Chinese exports and López and Rodil, 2008; Dussel and Trápaga, 2007; imports under a small number of chapters with respect Cárdenas and Dussel, 2011; Rodil and López, 2011; to trade with the United States reveals a trade pattern Neme, 2006) showed the growth in intra-industry trade dominated by a small group of sectors and China’s between Mexico and the outside world and particularly emergence in the nafta region, both as a supplier and with the United States. However, there is disagreement as a customer, but especially the former (see figure 5). concerning China. Some authors, including Neme More specifically, China’s trade with thenafta (2006), argue that, while real competition between countries is fairly concentrated under just five chapters. Mexico and China is obvious in some groups of products, In terms of exports, five chapters account for 63.0% of there is no clear winner and the Mexican exports Chinese exports to the United States, 59.8% to Canada to the United States market are not displaced by Chinese and 77.6% to Mexico. As regards Chinese imports, the exports. By contrast, other authors, including Cárdenas five main chapters represent 45.3% of those from the and Dussel (2011) and Dussel and Gallagher (2013), United States, 52.8% from Canada and 81.6% from hold that Mexico’s trade with the United States and Mexico (see tables 2 and 3). The trade balance in the China reflects weak integration with China and greater, five main chapters favours China over the United States albeit decreasing, integration with the United States. and Mexico, but in deficit vis-à-vis Canada.7 In terms Meanwhile, others (De la Cruz and Marín, 2011) of Chinese exports, China shows a clear surplus over all argue, on the basis of an analysis of causality, that the nafta countries in the five major chapters. In other statistical evidence points to a negative relationship words, China has a surplus in manufactures, but a deficit between Chinese and Mexican exports in the United in respect of natural resources and primary products. States market, and that Mexico must therefore try to While the United States is China’s main trading avoid displacement. partner, Mexico also experienced a significant relative Bearing in mind this situation, the present study increase in its relations with this Asian country. Some takes a different approach by exploring the phenomenon products are of particular importance in the relationship of intra-industry trade between China and Mexico in between Mexico and China, such as those under chapters the nafta region, complementing the previous analysis 84 and 85, which account for 49.4% of Chinese exports in order to gain a better understanding of trends at the to the nafta region. Meanwhile, Chinese imports from aggregate level. In fact, different patterns of trade are nafta show a greater dispersion with these two chapters studied in this work: North-North (United States with accounting for only 19%. Canada), North-South (United States and Canada with Mexico and China), and South-South (Mexico with China). In this article, we focus on these relations, always 7 Trade between China and Canada is not analysed in detail in this study. Suffice it to say that it is mostly inter-industry trade, based on taking account of the crucial role played by the United classic advantages relating to resourcing. In addition, its impact on States in the nafta region. Mexico’s trade is negligible.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 91

FIGURE 5 China: trade balance with the naftaa region, 1993-2011 (Millions of dollars)

600 000

500 000

400 000

300 000

200 000

100 000

0 1993 199 4 1995 199 6 199 7 1998 199 9 200 0 2001 200 2 2003 200 4 200 5 2006 200 7 200 8 2009 201 0 2011

Exports Imports Trade balance

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a North American Free Trade Agreement.

TABLE 2 China: products accounting for a larger proportion of exports to Canada, Mexico and the United States, 2011 (Annual average growth rates in percentages)

United States Canada Mexico Chapter Product 2011 1993-2011 2011 1993-2011 2011 1993-2011

84 Mechanical appliances, boilers and parts 23.9 25.7 19.8 28.9 24.8 29.6 85 Electrical machinery and equipment 23.8 18.6 24.3 22.7 43.4 38.5 95 Toys, games and sports requisites 5.7 9.7 5.6 13.3 2.5 26.9 94 Furniture, surgical equipment, not 5.4 17.6 5.7 20.0 … … specified elsewhere 64 Footwear, gaiters and the like parts 4.2 7.5 … … … … 62 Garments, not knitted … … 4.4 11.3 … … 90 Optical, photographic and medical devices … … … … 4.4 36.6 76 Aluminium and articles thereof … … … … 2.5 56.1 Selected total 63.0 16.5 59.8 20.0 77.6 33.8

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade).

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 92 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 3 China: products accounting for a larger proportion of imports from Canada, Mexico and the United States, 2011 (Annual average growth rates in percentages)

United States Canada Mexico Chapter Product 2011 1993-2011 2011 1993-2011 2011 1993-2011

84 Mechanical appliances, boilers and parts 11.8 11.2 … … … … 12 Seeds, oilseeds, grains and fruits 10.3 39.9 5.5 33.9 … … 85 Electrical machinery and equipment 9.7 14.4 … … 5.8 30.7 99 Non-specified products 7.0 28.3 … … … … 87 Land vehicles and parts 6.5 12.5 … … 14.9 ∞ 26 Ores, slag and ash … … 15.5 27.7 26.1 47 Wood pulp and derivatives … … 15.4 20.6 … … 44 Wood, articles of wood and charcoal … … 8.7 23.3 … … 27 Petroleum and petroleum products … … 7.8 26.8 22.4 ∞ 74 Copper and articles thereof … … … … 12.3 91.7 Selected total 45.3 15.3 52.8 24.3 81.6 51.3

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). ∞: the infinity symbol is used as the growth rate was calculated as starting from 0 in 1993.

As regards the intra-industry trade index (gl index) of China’s trade with the United States rose from 0.58 of China, the destabilizing effect of the trade balance in 1993 to 0.63 in 2011. In the case of China’s trade in bilateral relations with the nafta countries should with Mexico, the adjusted gl index dropped from 0.63 be taken into account by calculating the adjusted gl in 1993 to 0.41 in 2011. Lastly, the gl index of China’s index, as mentioned above (see figure 6). The gl index trade with Canada fell from 0.43 in 1993 to 0.29 in 2011.

FIGURE 6 China: intra-industry trade with the naftaa countries, 1993-2011 (Grubel-Lloyd index)

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0 1993 199 4 1995 199 6 199 7 1998 199 9 200 0 2001 200 2 2003 200 4 200 5 2006 200 7 200 8 2009 201 0 2011

United States Canada Mexico

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a North American Free Trade Agreement.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 93

From figure 6, we can draw at least three key (headings) shows more clearly the strong intra-industry conclusions. First, the levels of intra-industry trade character of Mexico’s relations with the United States found between China and the nafta countries are (its main trading partner) and China, as well as trends relatively low, especially compared with the patterns relating to the emergence of China in the nafta region. observed in developed economies (where intra-industry In particular, the calculation of the adjusted gl index trade generally accounts for 60% or 70% of total trade). for the bilateral trade between China, the United States This evidence shows that the emergence of China in and Mexico concerning chapter 84 shows a drastic the nafta region is linked to the global factory, but increase in intra-industry trade between China and without reaching full-industry integration. In fact, intra- Mexico (see figure 7). In fact, in recent years, bilateral industry trade is concentrated in just a couple of sectors. trade in these products (mechanical machinery and Second, China tends to record greater intra-industry equipment) between Mexico and China has reached top trade with the United States than Mexico or Canada. values of almost 1 (that is, nearly 100% intra-industry This illustrates the dominant role played by the United trade). The comparison with the trade between Mexico States as a global factory, which intensifies the intra- and the United States, which shows a clear stagnation industry trade flows with most of its trading partners. As and even a decline with regard to the beginning of the a result, the relations between its trading partners take period, raises new questions, in particular, whether on a secondary role, as is the case for South-South trade the intensification of intra-industry trade with China between China and Mexico and North-South trade between has halted the intra-industry specialization between China and Canada. Third, intra-industry trade between Mexico and the United States, which had taken off so China and the nafta countries has been trending vigorously in the late 1990s. The data seem to confirm downward over the last decade; only in the case of this hypothesis. the United States has this decline been contained. The data for the other important chapter for foreign This trend emphasizes the consolidation of China’s trade between Mexico and the United States and China, integration into the nafta region, pivoting around chapter 85 (electrical machinery and equipment), the central and dynamic role of the United States. In show a consolidation of its strong intra-industry nature contrast, China relates with the other two countries (90% in 2011). In contrast, the gl index for the other (Canada and Mexico) under a more complementary two bilateral trade relationships (Mexico with China strategy, although with intra-industry relations in certain and China with the United States) has stagnated (see specific sectors. figure 8). Nevertheless, the gl index of China with the Regarding trade between China and the United States, United States is higher than that of China with Mexico, there are 32 chapters with a gl index value above 0.5, owing to the leading role played by the United States in but these do not include the most dynamic chapters (84 intra-industry trade. and 85). However, if we analyse the adjusted gl index Something quite different occurs with the analysis taking into account the trade imbalance, these chapters of chapter 87 (motor vehicles), in which the strong do appear with high values. This suggests that most of intra-industry character of trade between Mexico and China’s trade with the United States is intra-industry the United States reappears, reaching maximum values trade, owing largely to the contribution of chapters during the last decade (see figure 9). A more irregular 84 and 85 (mechanical and electrical machinery and pattern is seen, however, for the other two bilateral trade equipment, respectively). Meanwhile, 18 chapters have relationships, with a surprising decline in the gl index a gl index over 0.5 in relation to trade between China at the end of the period. This trend clearly affects trade and Canada. However, between 1993 and 2011, the gl between Mexico and China, as well as trade between the index fell and the adjusted gl index was lower than 0.30, United States and China, reducing intra-industry trade which shows that most of China’s trade with Canada is flows to very low levels (close to 10%). Analysis with a complementary (inter-industry). higher level of disaggregation (four digits for vehicles, China’s trade with Mexico seems to be of an parts and accessories) suggests that the trends in the inter-industry nature, as only 11 chapters have a gl import and export of engines and vehicle accessories index exceeding 0.5. Moreover, the adjusted gl index are attributable to the lower labour costs obtained by decreased from 0.63 in 1993 to 0.41 in 2011, with a transnational corporations in China and to the rise of this sharp fall after 1999 (see figure 6). sector following the process of opening up the Chinese The detailed analysis of the most relevant chapters economy in the early 1970s (Cárdenas and Dussel, 2011; (84, 85 and 87), conducted at a more disaggregated level Dussel and Gallagher, 2013).

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 94 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 7 China, Mexico and the United States: intra-industry trade under chapter 84,a 1993-2011 (Grubel-Lloyd index)

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

China-Mexico China-United States Mexico-United States Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a Mechanical machinery and equipment.

FIGURE 8 China, Mexico and the United States: intra-industry trade under chapter 85,a 1993-2011 (Grubel-Lloyd index)

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

China-Mexico China-United States Mexico-United States Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a Electrical machinery and equipment.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 95

FIGURE 9 China, Mexico and the United States: intra-industry trade under chapter 87,a 1993-2011 (Grubel-Lloyd index)

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

China-Mexico China-United States Mexico-United States Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a Vehicles, parts and accessories.

Moreover, it is useful to compare the evolution of These chapters account for over half of Mexican United States imports from Mexico and China under exports to the United States and have a high rate of each of the relevant chapters (see figures 10, 11 and intra-industry trade, demonstrating Mexico’s integration 12). With the exception of chapter 87 (motor vehicles), into global value chains, especially with respect imports from China far exceed those from Mexico.8 to manufactures, which represent the output of the In fact, there seems to be a “substitution effect” for global factory. products originating in China, although in many cases, The same chapters (84, 85 and 87) of China’s imports from Mexico are not falling in absolute terms, exports to the United States are equivalent to 49.8% of which means that China is gaining a larger market share total Chinese exports, although the rate of intra-industry in other countries. trade is lower than in the case of Mexico. China is less The above finding does not apply to chapter 87 integrated, especially in relation to chapter 87, the area (motor vehicles) since, as shown in figure 12, Mexico where intra-industry trade between Mexico and the remains the top supplier to (and customer of) the United States is predominant. With respect to these United States in this production chain. The high level chapters, intra-industry trade is more prevalent between of intra-industry trade between these two countries in Mexico and the United States, while China’s trade with this sector is in part a reflection of the devolution of the the United States tends to be of a more complementary American automotive industry. This is clearly indicated nature (inter-industry trade). However, the exponential by Mexico’s share in the global trade of this industry: increase in intra-industry trade between China and Mexico rose from 2.4% of world exports and 0.6% of Mexico under chapter 84 (see figure 7), may be a result imports in 1993 to 5.3% of world exports and 2.7% of of a triangular trade process between China, Mexico imports in 2011 (wto, 2013). and the United States. Furthermore, given the wide variety of items included in these chapters, it is useful to analyse in 8 In some specific items this trend is even more dramatic, as in the detail what happens to the principal items they cover. case of the headings 8443, 8471, 8473, 8481 and 8525. Figure 13 shows the gl index of intra-industry trade

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 96 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 10 United States: imports from China and Mexico under chapter 84,a 1993-2011 (Billions of dollars)

120

100

80

60

40

20

0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Origin: Mexico Origin: China

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a Mechanical machinery and equipment.

FIGURE 11 United States: imports from China and Mexico under chapter 85,a 1993-2011 (Billions of dollars)

120

100

80

60

40

20

0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Origin: Mexico Origin: China

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a Electrical machinery and equipment.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 97

FIGURE 12 United States: imports from China and Mexico under chapter 87,a 1993-2011 (Billions of dollars)

50

45

40

35

30

25

20

15

10

5

0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Origin: Mexico Origin: China

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a Vehicles, parts and accessories.

for the top 13 items (at the headings level, four-digit engines), 8471 (automatic data processing machines, disaggregation) in Mexico’s trade with its main trading computers) and 8548 (electrical parts of machinery partner (the United States). Three bilateral trade pairings and apparatus). were analysed: Mexico-United States, Mexico-China and In addition to the above headings, China’s trade China-United States. The results show the prevalence with the nafta countries includes the following: of intra-industry trade between Mexico and the United 8443 (printing and auxiliary equipment), 8473 (parts States, exceeding that of the other bilateral pairings and accessories, except covers, for office machines), for 10 of the 13 items under consideration. These 8504 (electrical transformers, static converters and include headings 8536 (electrical switches, connectors), rectifiers), 8542 (integrated circuits and microstructures), 8708 (parts and accessories for motor vehicles) and 8711 (motorcycles and cycles fitted with an auxiliary 8408 (compression-ignition engines, diesel). The only motor) and 8712 (bicycles and other cycles, not motorized). three headings with a gl index that was lower for this China’s trade in these products involves very low levels pairing than for the other two bilateral trade pairings are of intra-industry trade, with flows predominantly in only headings 8409 (parts for spark-ignition internal combustion one direction.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 98 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 13 China, Mexico and the United States: intra-industry trade for selected headings in chapters 84, 85 and 87, 2011 (Grubel-Lloyd index)

8407 Spark-ignition internal combustion 1.00 8708 Parts and accesories 8408 Compression-ignition engines 0.90 motor vehicles (diesel, etc.) 0.80 0.70 8703 Motor vehicles for transport of 0.60 8409 Spare parts of spark persons (except buses) 0.50 ignition engines 0.40 0.30 8548 Electrical parts of 0.20 8471 Automatic data processing machinery and apparatus 0.10 machines (computers) 0.00

8544 Insulated wire and cable, 8517 Electric apparatus for line optical bre cable telephony, telegraphy

8536 Electric circuit switches 8523 Prepared unrecorded soundrecording media (non-photo) 8528 TV receivers, monitors 8525 Radio and TV transmitters, and cameras television cameras

Mexico-United States Mexico-China China-United States

Source: prepared by the authors, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). nes: not elsewhere specified. VI Conclusions

The results obtained in this study show the main trends equipment (chapter 85). These three types of products and implications for trade in the nafta countries arising account for more than 52.0% of Mexico’s total exports, from the emergence of China in North America. The which is evidence of the country’s growing intra-industry growing role of China over the last decade confirms specialization. The United States is Mexico’s primary its position as the leading supplier of the United States market, absorbing 82.2% of Mexican exports under market, pushing Mexico into second place. A similar these chapters. pattern has been seen regarding trade with Mexico, The fact that the United States’ intra-industry trade where China has established itself as the second largest under these chapters is likely to continue expanding, supplier, immediately after the United States. Similarly, and to a greater extent in relation to Mexico than China, China is Canada’s second largest supplier after the is no minor issue. Horizontal intra-industry trade is United States. the exchange of similar but differentiated products These trends are attributable to different factors, and vertical intra-industry trade refers to the transfer including, notably, intra-industry trade. In particular, of a product from one country to another at various intra-industry trade between Mexico and the United stages of development (intra-firm trade). Mexico has States grew robustly in the early years of nafta, with a the highest rate of intra-industry trade, as it integrates slight decline in 2000-2011, due to the 2001 recession in both modalities, while China’s intra-industry trade is the United States economy, and to China’s accession to basically horizontal, except under chapters 84 and 85. wto, which turned it into a major player in this region. The high levels of intra-industry trade with Mexico From a sectoral perspective, our results highlight the (chapters 84, 85 and 87) and China (chapters 84 growth in intra-industry trade relating to vehicles, parts and 85) show that they are part of the global factory whose and accessories (chapter 87), mechanical machinery and pivot is the United States. In Mexico, most of this trade equipment (chapter 84), and electrical machinery and is intra-firm, as in the case of the automotive industry, in

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. CEPAL REVIEW 114 • DECEMBER 2014 99 which Mexico has become a world power. Interestingly, inter-industry and in many cases linked to traditional however, none of the companies constituting that industry comparative advantages (resource endowment). In fact, is Mexican. Canada is the only nafta country with which China In summary, the findings suggest that intra-industry has a trade deficit. and intra-firm trade are closely connected in Mexico, China has become the world’s factory and enjoys a appearing to form two sides of the same process: Mexico’s trade surplus with all the nafta countries with respect participation in nafta and the progressive relocation of to its exports under the top four chapters (primarily the United States’ production industry. Nevertheless, manufactures). However, China posts a deficit in trade since China joined wto in 2001, nafta seems to have relating to natural resources (oil) and raw materials and been unable to create intra-area trade, except in the food (some of which are covered under its top five import automotive industry. As shown, China has staged a chapters). China has successfully penetrated nafta major breakthrough in the nafta region, despite having without any free trade agreements or any recognition achieved significant levels of intra-industry trade only as a market economy by the nafta countries. That free with the United States. This may be because joining wto trade area seems to work as a radiated wheel, in which gave China direct access to the United States market. United States acts as the hub, while China, Canada and By contrast, trade between China and Canada is mostly Mexico operate as the spokes.

Bibliography Balassa, B. (1963), “European integration: problems and issues”, Gazol, A. (2007), “Un nuevo tipo de proteccionismo (o el retorno American Economic Review, vol. 53, No. 2, Nashville, del permiso previo)”, Economía unam, vol. 4, No. 12, Mexico Tennessee, American Economic Association. City, National Autonomous University of Mexico. Cárdenas, H.L. and E. Dussel (2011), “El comercio intraindustrial Grossman, G.M. and E. Helpman (1990), “Trade, innovation and en México: Un comparativo entre China y Estados Unidos”, growth”, American Economic Review, vol. 80, No. 2, Nashville, Comercio Exterior, vol. 61, No. 4, July-August. Tennessee, American Economic Association. cefp (Centre of Public Finance Studies) (2006), Comentarios al Grubel, H.G. (1967), “Intra-industry specialization and the pattern of informe en materia arancelaria, 2005 (cefp/022/2006), trade”, Canadian Journal of Economics and Political Science, Mexico City. vol. 33, No. 3, Wiley. De la Cruz, J. and C. Marín (2011), “El impacto de China sobre Grubel, H.G. and P.J. Lloyd (1971), “The empirical measurement of América Latina en el mercado de Estados Unidos, un análisis intra-industry trade”, Economic Record, vol. 47, No. 4, Wiley. de causalidad”, Perfil de Coyuntura Económica, No. 18, Hamilton, C. and P. Kniest (1991), “Trade liberalization, structural Medellin, University of Antioquia. adjustment and intra-industry trade: a note”, Weltwirtschaftliches De la Cruz, J., J.A. Núñez and A. Ruiz-Porras (2008), “El impacto Archiv, vol. 127, No. 2. de la inversión extranjera directa de Estados Unidos en Helpman, E. (2006), “Trade, fdi, and the organization of firms”, América Latina y China: Evidencia empírica de causalidad”, Journal of Economic Literature, vol. 44, No. 3, Nashville, Globalización y regionalismo: Economía y sustentabilidad, Tennessee, American Economic Association. A. Ivanova and A. Guillén (coords.), Mexico City, Editorial jetro (Japan External Trade Organization) (2010), 2010 jetro Miguel Ángel Porrúa. Global Trade and Investment Report. A Global Strategy for Dornbusch, R. (1992), “En defensa de la apertura comercial en Japanese Companies to Open New Frontiers in Overseas los países en desarrollo”, Análisis Económico, vol. 10, Markets, Tokyo. No. 20, Mexico City, Metropolitan Autonomous University- Krugman, P. (1995), “Growing world trade: causes and consequences”, Azcapotzalco. Brookings Papers on Economic Activity, vol. 26, No. 1, Dussel, E. and K.P. Gallagher (2013), “nafta’s univited guest: Washington, D.C., The Brookings Institution. China and the disintegration of Northa American trade”, cepal López, J.A. and O. Rodil (2008), “Comercio intrandustrial e intrafirma Review, No. 110 (LC/G.2572-P), Santiago, Chile. en México en el contexto del proceso de integración de América Dussel, E. and A. León González (2001), “El comercio intraindustrial del Norte (1993-2006)”, Economía unam, vol. 5, No. 13, en México, 1990-1999”, Comercio Exterior, vol. 51, No. 7, Mexico City, National Autonomous University of Mexico. Mexico City, July. Navaretti, G.B., J.I. Haaland and A. Venables (2002), Multinational Dussel, E. and Y. Trápaga (coord.) (2007), China y México. Corporations and Global Production Networks: the Implicaciones de una nueva relación, Mexico City, Jornada Implications for Trade Policy, London, Centre for Economic Ediciones. Policy Research. Feenstra, R. and H. Looi Kee (2009), “Trade liberalization and export Neme, O. (2006), La competencia entre México y China: La disputa por el variety: a comparison of Mexico and China”, China’s and mercado de Estados Unidos, Mexico City, National Autonomous India’s Challenge to Latin America. Opportunity or Threat?, University of Mexico (unam)/Miguel Ángel Porrúa. D. Lederman, M. Olarreaga and G. Perry (eds.), Washington, oecd (Organization for Economic Cooperation and Development) D.C., World Bank. (2002), “Intra-industry and intrafirm trade and the internalization Gaceta Parlamentaria (2012), “Informes sobre el uso de la facultad of production”, oecd Economic Outlook, No. 71, Paris. conferida al Ejecutivo federal en materia arancelaria, que se Rodil, O. and J.A. López (2011), “Efectos del Tratado de Libre presenta de conformidad con el artículo 131 de la Constitución Comercio de América del Norte sobre el comercio de México: Política de los Estados Unidos Mexicanos”, year 16, No. 3662-D, Creación de comercio y especialización intraindustrial”, Revista Palacio Legislativo de San Lázaro, 7 December. de Economía Mundial, vol. 27.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. 100 CEPAL REVIEW 114 • DECEMBER 2014

Rodríguez-Pose, A. and G. Petrakos (2004), “Integración económica Viner, J. (1950), The Customs Union Issue, New York, Carnegie y desequilibrios territoriales en la Unión Europea”, eure, Endowment for International Peace. Revista Latinoamericana de Estudios Urbanos Regionales, Wang, J. (2004), “China’s Changing Role in Asia”, Washington, vol. 30, No. 89, Santiago, Chile, May. D.C., Atlantic Council [online] http://www.acus.org/docs/ Rosales, O. and M. Kuwayama (2012), China and Latin America 0401China_Changing_Role_Asia.pdf. and the Caribbean: Building a Strategic Economic and Trade wto (World Trade Organization) (2013), “International trade Relationship, Libros de la cepal, No. 114 (LC/G.2519-P), statistics, 2012” [online] http://stat.wto.org/StatisticalProgram/ Santiago, Chile, Economic Commission for Latin America and WSDBViewData.aspx?Language=E. the Caribbean (eclac), March. United Nations publication, Yu, Ch., L.X. Xue and S. Hong (2006), “China y los acuerdos Sales No. E.12.II.G.2. de libre comercio”, Revista del cei. Comercio Exterior e Verdoorn, P.J. (1960), “The intra-bloc trade of Benelux”, Economic Integración, No. 7, Buenos Aires, International Economics Consequences of the Size of Nations, E.A.G. Robinson (ed.), Centre (cei), December. New York, Macmillan.

THE IMPACT OF CHINA’S INCURSION INTO THE NORTH AMERICAN FREE TRADE AGREEMENT (nafta) ON INTRA-INDUSTRY TRADE • JORGE ALBERTO LÓPEZ A., ÓSCAR RODIL M. AND SAÚL VALDEZ G. Work, family and public policy changes in Latin America: Equity, maternalism and co-responsibility

Merike Blofield and Juliana Martínez F.

ABSTRACT Taking account of the substantial increase in female labour market participation that has occurred throughout the Latin American region, this article describes policies adopted with the aim of reconciling work and family responsibilities between 2003 and 2013, and the implications of their design for socioeconomic and gender equity. We look at the cases of Argentina, Brazil, Chile, Costa Rica and Uruguay, five countries which, on the basis of their track records, are the best placed to implement policies to reorganize time, income and services. The empirical analysis indicates, first, that these changes have contributed to socioeconomic equity more consistently than to gender equity. Second, the scale and type of change was found to vary significantly from one country to another. The article concludes by raising a number of substantive questions about the measures, their implementation and effectiveness, and the variations between countries.

KEYWORDS Women, women’s employment, family, gender roles, gender equality, women’s rights, social policy, Argentina, Brazil, Chile, Costa Rica, Uruguay

JEL CLASSIFICATION I38

AUTHORS Merike Blofield is an associate professor at Miami University. [email protected] Juliana Martínez F. is an associate professor at Costa Rica University. [email protected] 102 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

Latin America is undergoing a “silent revolution” (ilo/undp, 2009), women’s absence from the labour (Goldin, 2006) at the intersection between work and market worsens poverty and social inequality. Second, family. Seven of every 10 women of childbearing age women with income of their own are overrepresented are in the workforce and increasingly live in female- among the self-employed and in domestic work; in the headed —and often single-parent— households. Yet care latter occupation three quarters of individuals, almost responsibilities continue to be shouldered primarily by all of them women, had no pension provisions in 2008 women (ilo/undp, 2009; Sojo 2011; Montaño 2010; (ilo, 2011). Women in the labour market thus have less eclac, 2013a). The tensions in gender relations resulting protection than their male peers —in urban areas only from this combination of change and continuity in work 36% of women have social security provision, compared and family are set against a backdrop of deep social with 49% of men (ilo/undp, 2009)—. Third, those inequality (Cornia, 2010; eclac, 2011; López-Calva and who shoulder heavier care burdens have less social Lustig, 2010). What have governments done to address protection: in 12 Latin American countries, women these tensions and with what outcomes for inequality? with incomplete primary schooling have between 2 This article explores that question on the basis of policies and 3.5 more children than those who have secondary adopted during the 10 years of economic expansion schooling or more (eclac, 2011, p. 85). In female- since 2003 in the five countries in which —as will be headed single-parent households, the double burden of discussed later— social policy is, relatively speaking, singlehandedly providing both care and income worsens most developed in the region: Argentina, Brazil, Chile, the negative consequences of labour informality and lack of Costa Rica and Uruguay. social protection. It is worth first discussing the social and policy The situation today enshrines both challenges relevance of addressing tensions between work and and opportunities (ilo/undp, 2009; Sojo, 2011; family responsibilities. First, the responsibility for care Montaño Virreira, 2010; Chioda, 2011). With the right work represents a great barrier to women in terms of public policies, governments could break the cycle accessing the labour market (eclac, 2010). In such a highly of inequality and promote more inclusive social and unequal region as Latin America, this particular gender economic development. Some measures can achieve gap is heavily marked by socioeconomic stratification. more than one objective at a time. For example, it is The gap in female labour market participation between universally agreed that early education and care services the richest and poorest quintiles is 30% on average develop human capital and promote equal opportunities. and has not shrunk since 1990, despite the large rise in Combined with suitable opening hours and schedules, women’s overall labour market participation.1 Given that these services could also support working mothers and the odds of overcoming poverty are proportional to the fathers. Conversely, without adequate responses to the odds of having more than one income in the household tensions between work and family responsibilities, socioeconomic and gender inequalities will continue to deepen and work against equity and development in The authors are grateful to Ariel Armony, Eleanor Faur, Janet Gornick, the region. Evelyne Huber, Beatriz Magaloni, Esther Mancebo, Carmen Midaglia, How much progress has been made in adopting public Jennifer Pribble, Soledad Salvador, Silke Staab, Juan Diego Trejos, policies to reconcile work and family responsibilities and two anonymous readers at the cepal Review. They also thank María Cristina Alcántara, Felipe Sterquino Itaborai and, especially, in those countries that are in the best position to do so? Diana León for their rigorous and enthusiastic research assistance. Section II sets forth an analytical framework to examine Any error of assessment is the sole responsibility of the authors. these changes. Section III explains the methodology 1 For example, in Brazil, the gap is almost 40%, with only 46.5% and section IV presents the empirical findings. Lastly, of women in the lower-income bracket in the labour force versus 84.4% of women in the higher income quintiles (pnad/ibge, 2014; section V concludes with a summary and discussion of Sorj and Fontes, 2012). the findings.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 103

II Analytical framework2

A broad universe of public policies affect the relationship leave) or last hours within a single work day or between work and family responsibilities, whether by week (as in part-time or flextime measures). With action or by omission, ranging from urban planning and sequential policies, care remains within the family, public transportation, to social policy per se on issues performed historically by women, although under of leave and care (Monge, 2006). Below, we distinguish more recent measures family responsibilities have between different types of social policy that affect increasingly been extended to men through paternity work-family tensions,3 then discuss how these maintain and parental leave.5 or change socioeconomic and gender relations. The • Policies that defamilialize care refer to transfers and section concludes by presenting examples of specific services that shift the responsibility for care provision policies aimed at changing initial socioeconomic and from families —and specifically from women— to gender inequality. services with some degree of State involvement. These may be provided directly by the State, involve 1. Policies that reconcile work and family incentives or subsidies for private provision, or mandate employer provision of subsidies, services Whether or not they ultimately have positive or negative or both. As with sequential policies, these measures impacts from the point of view of socioeconomic and often revolve around mothers and female workers gender equality, measures aimed at reconciling work but —as detailed below—have increasingly begun and family responsibilities can function in three ways: to make men and fathers eligible as well in Latin by reallocating time, income or services. Specifically, America (ilo/undp, 2009). they enable individuals to alternate time spent at work • The third category addresses regulations on the with time spent on caregiving within the family (Durán, hiring of home-based paid care providers specifically, 2004), transfer family care to services partly provided where the household is thus turned into a workplace. by the State, or regulate the purchase of private care by These are “regulatory policies on home-based families. These types of State intervention are sequential, care services”. The one-to-one hiring of mostly “defamilializing” (Martínez Franzoni, 2008; Orloff, 2009) female personnel for work in familial settings is and regulatory, respectively,4 and often addressing work- an alternative to provision in institutional settings, family tensions is not in fact their primary purpose; this is and has different implications for the State’s role the case with preschool childcare services, for example. in measures that defamilialize or commercialize In addition, as will be illustrated below, each of these care, since paid domestic work is not usually types of measure can be addressed from the standpoint included in policies for reconciling paid work and of either labour policy or social policy. family responsibilities. • Sequential policies refer to measures that support The main focus of our study regarding the third income during periods —months, weeks or days— category is on domestic work that is considered spent on caregiving. They include maternity, “unskilled”,6 although the analysis can equally be paternity or parental leave, as well as flextime extended to skilled occupations, such as nursing. As and part-time work policies. Sequencing can last with other care occupations, home-based care tends to months and involve many work days (as in maternity

5 Parental leaves were established first in Europe, specifically in 2 This analytical framework is developed in greater detailed in Blofield Sweden, in 1974. Paternity leaves came later and were created to and Martínez Franzoni (2014). encourage men to use parental leave. Both maternity leaves and 3 This article refers to work-family tension indistinctly as work-family, parental and paternity leave are considerably longer in Europe and in and work and family responsibilities. English-speaking countries, except for the United States, which lags 4 A first approach to changes in these three types of measures in Latin behind even by comparison with the poorer Latin American countries. America (although with a different definition of regulatory measures) 6 By “unskilled” we mean that these employees rarely have formal was developed by Martínez Franzoni and Camacho (2006 and 2007) training for their occupations, although they provide services that on the basis of an adaptation of Durán (2004). require a broad range of practical but socially undervalued skills.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 104 CEPAL REVIEW 114 • DECEMBER 2014 be overwhelmingly female and penalized by lower pay unpaid, poorly paid or voluntary work, or by promoting than that received by workers in comparably skilled the provision of services through trained professionals occupations. This penalty derives from three factors: care hired under formal conditions of employment. occupations have historically been seen as extensions of Sequential, defamilializing and regulatory policies “naturally” female roles; they are perceived as intrinsically are all qualitatively distinct and complementary in work- rewarding; and, as “sacred activities”, they are “above” family reconciliation. For this reason, more policy in financial recognition (England and Folbre, 1999). one dimension cannot be expected to make up for less Insofar as caregiving (whether paid or unpaid) policy in another dimension. Empirical analysis must involves an emotional connection between caregivers and therefore consider them simultaneously. those who are being cared for, labour market regulations regarding these occupations are critical to the status of 2. Implications for socioeconomic and the care providers as workers and to the type of service gender equity performed (see for example: Folbre, 19957 and Williams, 2010). What makes these occupations distinct is that the Depending on how they reallocate time, income workplace and the household overlap and that the bonds support and services, whether by design or through between caregivers and care-receivers tend to be more implementation, sequential, defamilializing, and regulatory personal, challenging the regulation of these occupations measures can entrench or alter socioeconomic and gender in general, and those based in the employer’s home in inequality —as produced by the labour market and the particular. With caregiving historically associated with sexual division of labour between women and men—. a servant culture in many countries, discrimination This work does not evaluate how these measures change has been rooted in laws and labour codes, with long the social structure —if indeed they do— but focuses working hours and very limited labour protections rather on their design. and benefits. Let us start by looking at gender inequality. Domestic or home-based work is, precisely, one The literature on welfare states and gender relations of the main ways of resolving the tensions between distinguishes “maternalist” policies from those that family and work in Latin America: around 15% of the promote “social co-responsibility” in caregiving (ilo/ economically active female population is employed in undp, 2009). Maternalist measures (Orloff, 2006) domestic services, with a similar percentage of households recognize the importance of caregiving and exalt as employers (eclac, 2013b; ilo 2012, pp. 59-60). “women’s capacity to mother” (Koven and Michel, Weak regulation of this occupation by the State has, 1993, in Orloff, 2006, p. 4), and thus reward women as in practice, allowed high-income families to reconcile the primary and main providers of care.8 Here, there is work and family responsibilities at the expense of such a social recognition of care, but the gender gap in its reconciliation by these female workers (Blofield, 2012). provision remains intact. Extended maternity leave, Another, more collective form of home-based tax incentives or monetary transfers for stay-at-home care (also referred to as family day care) is care in the mothers are examples of maternalist measures. They are caregiver’s household. While such care has no doubt used as alternatives to greater participation by fathers, existed informally throughout time, in the past few the use of services, or both. decades it has become an object of government funding We distinguish maternalist policies from policies and regulation as part of social service extension to that establish a “maternalist floor”, acknowledging low-income families. the specific role of women in gestation, giving birth, There are examples of all three types of measure breastfeeding and establishing early routines and bonds. in both labour and social policy. In all three cases, the These are analytical categories, but also thresholds that State may have an impact by action or by omission, leaving the solutions to work-family tensions to family strategies and informal solutions by women, through 8 Historically, maternalist movements “made arguments for gender justice: women should be recognized and compensated by the State for their unique contribution to society —through maternity and childrearing” (Orloff, 2006, p. 10)—. Maternalist arguments can lose 7 Folbre (1995) defines care work, whether paid or unpaid, as work ground to a view of care that involves but transcends women. Orloff that involves connecting to other people and trying to help people meet argues that in European and North American countries, maternalist their needs. Folbre argues that the intrinsic motivation involved in care claims have lost elite and popular support over time, although not labour as something people do for a third party poses challenges for always in the direction of more gender equity but rather in the direction markets to organize and pay for care work. of less social equity.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 105 have changed over time. For example, the demarcation knowledge of their effects in terms of reducing gender of what the International Labour Organization (ilo) inequality is still very incipient. considered a minimum maternity leave —and which is In terms of socioeconomic equity we follow Esping- taken here to define what we call the maternalist floor— Andersen’s (1990) distinctions between eligibility based has changed over time. In 1952, ilo set maternity leave on needs, contribution, or citizenship. The implications at 12 weeks in Convention 103, but by 2000 this had of these criteria vary depending on the point of departure: increased to 14 weeks (Convention 183). relative social equality, low poverty rates and basically A maternalist floor is essential to protect women formal labour markets, or the opposite, as in Latin in their role as mothers. Maternalist policies, however, America, where highly informal labour markets limit can affect equity in contradictory ways. Although these the scope of contribution. Here eligibility mechanisms policies publicly recognize and support motherhood must be more heavily based on citizenship and need as a central dimension of women’s lives and can thus than in developed countries.9 elevate the status of mothers, they can also reinforce the Furthermore, it may be observed that policies based notion that care is women’s sole responsibility. It is an on contributions through formal employment tend to empirical matter to establish which maternalist policies restrict benefits to those making regular contributions also promote gender equity by levelling the playing field as well as their dependents —often even segmenting rather than reinforcing the sexual division of labour. benefits between the former and the latter—. This mode In contrast to maternalism, co-responsibility of eligibility applies primarily to middle- and upper- redistributes care responsibilities, both from families income groups of the population, therefore reinforcing to the State (State co-responsibility) and from women socioeconomic inequities.10 These policies also tend to to men (paternal co-responsibility). reinforce inequities between groups within the formal State co-responsibility in work-family reconciliation labour force; for example, workers on fixed-term contracts policies not only produces defamilialization through and paid domestic workers may not legally have access publicly provided or subsidized provision of early to the same rights. childhood education and care (ecec), but also, and very Taking into account the prominence of informal importantly, requires the provision of full-time care, thus relations in Latin American labour markets (in temporary allowing for the use of services by working parents. and domestic work, for example), the question is whether Public social investment in early childhood education policies extend protections beyond formal employment and care services that do not correspond to a typical to a broader scope of wage workers, self-employed full-time workday, may promote state co-responsibility workers, or both.11 We also assess whether policies move in child education but not in work-family reconciliation. beyond an individual’s formal labour status to a basis Paternal co-responsibility policies promote sharing in citizenship or need. We thus evaluate whether work- of caregiving by incentivizing fathers’ involvement. This family policies —whether sequential, defamilializing or makes it possible to alternate income-generating and regulatory, and maternalist or pro-co-responsibility— alter caregiving responsibilities between women and men or reproduce the initial, labour-market stratification and (Fraser, 1997). This “feminization of the male life cycle” therefore also enhance socioeconomic equity.12 (Esping-Andersen, 2009, p. 99) generally takes the form Second, informality pervades care arrangements, of sequential policies that enable the reorganization either through the informal hiring of domestic workers of gender roles between mothers and fathers, without or through female unpaid work. Without government undermining pay or job continuity. By shifting more responsibility for care to the State and to fathers, co-responsibility policies have the 9 Nevertheless, contributory social protection has more relative capacity to reduce gender inequities in the burden of care. coverage and social investment than non-contributory programmes. The extent to which they actually do so is a matter for 10 For a discussion of the relevant cases where universalism has been built around contributory policy, see Martínez Franzoni and empirical analysis. The Swedish experience shows that Sánchez-Ancochea (2013). paternity leave which is not transferable to the mother 11 In Latin America labour and social protections reach workers in promotes fathers’ involvement in childcare more than various degrees, so that there is a continuum from the most formal to does parental leave, for which both fathers and mothers the most informal arrangements among wage workers as well as the self-employed. Rather than making up an informal “sector”, informality are eligible. Both paternity leave, consisting of a few thus cuts across the labour market. days to support the mother, and parental leave, have 12 See Pribble (2013) for a broader discussion of equity-enhancing been established very recently in Chile and Uruguay, so social policies.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 106 CEPAL REVIEW 114 • DECEMBER 2014 intervention, the care burden tends to fall particularly of inequality under another; we must instead examine their heavily on low-income women. In other words, degrees interaction across distinct policy initiatives. The three types of familialization in care (Martínez Franzoni, 2008; Orloff, of policy —sequential, defamilializing, and regulatory— 2009) differ by income levels. The higher the family’s may have various implications for socioeconomic and income, the greater the capacity to transfer much of the gender stratification. Empirically speaking, in terms of burden of domestic work to lower-income women hired gender equity we attempt to establish whether specific for that purpose and avoid women having to negotiate the measures promote a maternalist floor, maternalism, or sharing of care responsibilities with their male partners. co-responsibility. Regarding socioeconomic equity, Conversely, the lower the household income, the fewer we address whether or not protection extends beyond the options to outsource domestic and care work. wage work and contributory mechanisms to cover own- Given the interactions between socioeconomic and account or temporary workers, or on the basis of need gender inequalities, it would be wrong to subsume one type or citizenship.

III Methodology

Below, we analyse types of paid leave for birth, full-time the gap calls for more comprehensive assessments that childcare services and the levelling of the rights of paid include the degree of implementation of the changes domestic workers with the rest of the workforce. These in question. Legislation and executive measures that policies do not, of course, represent the entire universe have been discussed but not adopted are not included, of sequential, defamilializing and regulatory policies, nor are collective agreements (which are important in but are nevertheless emblematic. Uruguay, Brazil and Argentina, in particular) or state-level The cases examined correspond to five Latin legislation —which is particularly prominent in the two American countries that are in the best position, relatively federal countries analysed (Argentina and Brazil)— that speaking, to respond to changes in families and labour goes beyond the federal minimum. markets in the region: Argentina, Brazil, Chile, Costa The empirical evidence used comes from primary Rica and Uruguay. This is the group of countries with and secondary sources. The first include laws, executive modest social gaps (Filgueira, 2011), comparatively decrees and policy papers, executive committee or formal labour markets and greater social investment. congressional reports and interviews, as well as press Accordingly, they are generally considered to have the articles. Secondary sources include country analyses, same social policy regime-Statist (Martínez Franzoni, several of which take a sociological approach or are 2008) or advanced (Huber and Stephens, 2012). They intended to provide assessments or policy inputs.13 have highly institutionalized political systems and greater State capacity (Pribble, 2013), and have reached a more advanced stage of the demographic transition 13 There is an interesting and growing body of literature on the subject (eclac, 2010). for the five countries. For example, for Argentina, see Faur (2011); We compare federal policies in place in 2003 Faur, Esquivel and Jelin (2012); Gherardi, Pautassi and Zibecchi (2012) and Rodríguez and Pautassi (2014); for Brazil, see Hirata and Araujo and 2013 across these countries and over time. It is Gimarães (2012), and Sorj (2013); for Chile, see Bentancor and De important to be aware that a large gap can exist between Martini (2012), and Staab (2012) (the latter with an explicit focus measures formally in place and their implementation. on policymaking); for Costa Rica, see Román and Morales (2010); Sauma (2012); for Uruguay, see Aguirre and Ferrari (2014) —focusing Nevertheless, the adoption of these measures is in itself strictly on policy adoption—; Batthyany, Genta and Perrotta (2012), indicative of existing policy priorities, and understanding and Salvador (2013).

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 107

IV Empirical evidence: work-family policies in 2003 and 2013

Below, we examine the empirical evidence on paid leave for Argentina, in all the countries women continue to for childbirth, early childhood care and education and the accumulate pension entitlements during maternity leave. rights of paid domestic workers, as cases of sequential, Figure 1 distinguishes between measures defamilializing and regulatory policies, respectively. The corresponding to the maternalist floor (equivalent to the findings show that more changes have occurred in care ilo threshold of 14 weeks, shown by a line), maternalist services and the regulation of paid domestic work than measures (above the line), and measures that promote in leave associated with paid employment. co-responsibility either through non-transferable leave The broadest legal changes in this last category have for fathers (paternity leave), or through leaves that are occurred in Chile and Uruguay, and affect both eligibility shareable between mothers and fathers (parental leave). and length of leave. In the other three countries, groups In 2003 maternity leaves were at least the length of of female workers have gained eligibility for maternity the maternalist floor, except in Uruguay and Argentina. leave, either through legal rulings (Brazil and Costa Uruguay provided two weeks less, and Argentina —with Rica) or legislative reform (Argentina). a leave of 90 days— slightly more than a week under Early childhood care and education services have the 14 weeks established by ilo. undergone changes to varying extents in all the countries. During this period, leaves changed considerably in In general, more progress has been made in promoting Chile and Uruguay. Already by 2003 Chile had the longest socioeconomic equity than in paternal co-responsibility. maternity leave of the five countries and the reform of 2011 (Law 20.545) added a further three months. That 1. Sequential measures: increased maternalism reform also extended paternity leave (from 3 days to 5) and socioeconomic equity and allowed fathers to use up to half of the final three months of postnatal leave. The reform accordingly How much have employment-based paid leaves changed promoted a combination of marked maternalism with and how? Figure 1 shows the duration in weeks of paid timid progress in paternal co-responsibility.15 maternity, paternity and parental leaves in the five countries In Uruguay various changes were made to the leave in 2003 and 2013. All the leaves analysed correspond scheme and November 2013 brought a comprehensive to a full salary —although sometimes with a ceiling, reform for the private sector (Law 19.161). As a result, as in Chile— but they differ in the source of financing in the private sector maternity leave increased from 12 (from social security in Argentina, Brazil and Uruguay; to 14 weeks. Paternity leave was extended as well, from from a combination of social security and employers in 3 consecutive days in the private sector and 10 working Costa Rica; and from the national budget in Chile). The days in the public sector to a further 10 consecutive best scenario for socioeconomic and gender equity is days in the private sector, for both wage workers and when leaves do not carry a direct cost for employers.14 independent workers paying into Uruguay’s Social Paid maternity leave was established in the countries’ Security Bank (Banco de Previsión Social, bps).16 The labour codes in earlier decades, and by 2003 ranged extension took effect immediately for maternity leave, from 12 weeks in Uruguay to about 18 weeks in Chile. but more gradually for paternity and parental leave. In 2003, paternity leave was zero in Costa Rica and Previously, public sector paternity leave had been Uruguay, 1 day in Chile, 2 days in Argentina and 5 days 3 days since the reform of 1989 (Law 16.109) and 10 days in Brazil (enshrined in the Constitution of 1988). Except from 2008 (Law 17.930), when private sector workers also gained another 3 days of paternity leave. To the

14 Leaves that do represent a direct cost for employers not only 15 discourage the hiring of women, but also affect smaller firms more See Lupica (2013) for an analysis of the process by which the and so are regressive in terms of the production structure. For data proposal was prepared. and analysis on this point for the region overall, see Pautassi and 16 Self-employed in the formal sector who pay social contributions Rico (2011). through profession-related funds are not included.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 108 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 1 Leave for childbirth and gender equity by country, 2003 and 2013 (Weeks)

35

30

25

20

15

10

5

0 Chile 2003 Chile 2013 Brazil 2003 Brazil 2013 Uruguay 2003 Uruguay 2013 Argentina 2003 Argentina 2013 Costa Rica 2003 Costa Rica 2013

Maternity leave Paternity leave Shareable parental leave Maternalist oor Source: prepared by the authors, on the basis of legislation in the respective countries and data from R. Ray, J. Gornick and J. Schmitt, “Who cares? Assessing generosity and gender equality in parental leave policy designs in 21 countries”, Journal of European Social Policy, vol. 20, No. 3, Sage, 2010. Note: the duration of paternity leave was registered in relation to a unit of a 7-day week. For Uruguay, paternity leave was estimated on the basis of an 8-week maternity leave (Law 19.161, article 2).

3 consecutive days financed by the employer, the 2013 had yet to reach the ilo standard maternalist floor in reform added 3 more days in 2014 financed by the 2014. What is more, apart from the cases of Chile and Social Security Bank, increasing to 7 more in 2015 and Uruguay, and despite draft legislation aimed at extending 10 more in 2016. Accordingly, the 13 consecutive days paternity leave to between 2 and 4 weeks in Argentina, in the private sector are more or less equivalent to the Brazil and Costa Rica, paternity leaves continue to working days of leave in the public sector (depending be minimal. on whether public holidays occur during the leave).17 Given the prevalence of labour informality, In 2016, parental leave will allow the mother or employment-based leaves have been particularly unequal the father to work a half-day after the 8-week maternity from a socioeconomic perspective. Since maternity leave leave is over, until the child is six months old. This is by far the more substantial, figure 2 shows eligibility part-time leave will last for up to 4 months or, for criteria in each country in 2003 and 2013. Specifically, comparison purposes, 8 weeks full-time, as shown in it shows whether only some female wage workers, all figure 1 (following Ray, Gornick and Schmitt, 2010). female wage workers or all female workers, including The reforms adopted in Chile and Uruguay have the self-employed, are eligible. different implications for gender equity. Maternity leave In 2003, in Chile and Argentina only some female was extended beyond the maternalist floor in Chile, wage workers were eligible for maternity leave, leaving and brought up to that threshold in the private sector in out precisely those who were socioeconomically more Uruguay. Paternity leave is twice as long in Uruguay as vulnerable. In Chile only those with employment in Chile, and in Uruguay it is financed mainly by social contracts were eligible, which were mostly those in the security and to a lesser extent by the employer, while in upper income quintiles (Pribble, 2006, p. 90), although Chile and Brazil it is financed by the employer. In sum, domestic workers with contracts were included in 1998. the change favours parental co-responsibility more in Domestic workers were excluded in Argentina (where Uruguay than in Chile, in principle. At the other extreme, they represent 17% of the female urban workforce). In Argentina was the only one of the five countries that Brazil, Costa Rica and Uruguay all female wage workers were eligible, including, in the case of Brazil, independent and rural seasonal workers. This last group does not 17 See Salvador (2013) for an estimate of the costs of the reform in need to show social security contributions to qualify for Uruguay, which is also methodologically useful for the other countries. maternity leave with benefits equivalent to the minimum

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 109

FIGURE 2 Maternity leaves and socioeconomic equity: eligibility by country, 2003 and 2013 (By labour status)

3

2

1

0 Chile 2003 Chile 2013 Brazil 2003 Brazil 2013 Uruguay 2003 Uruguay 2013 Argentina 2003 Argentina 2013 Costa Rica 2003 Costa Rica 2013

Some female wage workers All female wage workers Own-account workers

Source: prepared by the authors, on the basis of legislation in each country.

wage. Since 2003, all five countries have changed the the legislature extended maternity leave to some insured eligibility rules towards greater socioeconomic equity; self-employed women in 2013.20 in Costa Rica and Brazil through legal action and in the Informal female workers who do not contribute other three countries by legislative reform. to social security —generally a large proportion of In Argentina, a legal reform adopted in 2013 (with the lower-income quintiles— are excluded from these implementing regulations adopted in April 2014) extended eligibility criteria, except for rural workers in Brazil.21 maternity leave to domestic workers, making all female wage workers eligible.18 In Brazil, a 2012 ruling by the 2. Defamilializing measures: advances towards supreme labour court extended the right to maternity State co-responsibility with socioeconomic leave to workers on temporary contracts. In Chile, the equity in early childhood care and education 2011 parental leave reform extended maternity leave to all female wage workers, seasonal and independent Below we examine national policies on early childhood workers. Costa Rica made insurance mandatory for care and education, principally for children aged 0 independent workers as part of a deal on pensions between to 3 years, and tangentially for those aged between the government, chambers of commerce, unions and 4 and 5 —the two age groups targeted by early childhood other organizations.19 Although the agreement initially care and education services and preschool education—. included only the services sector, in 2004 a ruling by the Depending on their opening hours and coverage of low- Constitutional Court extended it to monetary transfers income sectors, care services can promote both State and, accordingly, to maternity leave. Seasonal workers co-responsibility and socioeconomic equity. are considered wage workers, and hence their insurance Preschool education tends to reflect concerns of is mandatory. However, there have been obstacles to human capital formation more than early childhood care adapting the conditions of insurance, for example, for seasonal rather than monthly contributions. In Uruguay, 20 As pointed out by Soledad Salvador (2014), in fact all benefits (paternity and parental) are extended to men and women who pay into the Banco de Previsión Social (not independent funds such as profession-related funds), business-owners with up to one dependent, 18 Law 26.844 with implementing regulations in executive decree and the self-employed. Those with dependent workers are reported 467 of 2014. as employers. 19 In 2008, a pension reform recognized and provided compensation 21 Women in this position can be eligible for other cash transfers for time mothers devote to raising children, increasing their probability as mothers rather than as workers, mainly through conditional cash of eligibility for an old-age pension in the future. transfer schemes for children, pregnant women or both.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 110 CEPAL REVIEW 114 • DECEMBER 2014 services. Services for children aged 0-3 give a closer and education services for children aged 0-3 years, and picture of governments’ commitment to work-family whether they promote socioeconomic equity. balance —i.e. State co-responsibility— through services With the exception of Argentina, in 2003 the countries that partially defamilialize care and provide full-time included here all had national programmes for early service coverage. This is also the age group for which childhood care and education for children aged 0-3. In there is greatest social reticence regarding mothers Brazil, Chile and Uruguay these programmes —which combining motherhood and paid work and, in general, will be discussed in more detail later (childcare centres, any outsourcing of care outside the household. National Board for Nursery Schools (junji), Fundación Argentina, Brazil and Chile have had legislation Integra and Child and Family Assistance Centres (caif), obliging large firms to provide some childcare arrangements respectively)— had explicit eligibility criteria. By contrast, since before 2003.22 These arrangements depend on the in Costa Rica children were eligible for Education and number of female workers in the firm, are limited to Nutrition Centres/Integrated Centres for Childcare mothers (fathers are not eligible) and were created to (cen-cinai) on the basis of economic need, but this allow for breastfeeding during the legally established criterion was applied fairly discretionally and varied period. In Chile, employers with 20 or more female depending on the demand at each centre, becoming more workers must provide a day-care service for children targeted where demand was heavier. Brazil and Uruguay under age 2. In Brazil, companies with 30 or more had set coverage targets, quite ambitious ones in the female workers must provide day care for babies up to first case. In socioeconomic terms, eligibility criteria 6 months old (i.e. for two months, between the end of the for early childhood care services varied widely from no obligatory maternity leave and the end of the six-month standard criteria in Costa Rica to universal eligibility in breastfeeding entitlement). In Argentina, employers with Brazil based on the 1988 Constitution. 50 or more female workers have been legally obliged By 2013, Argentina had established a national to provide a day-care service since 1970. However, programme of Child Development Centres (cedis) under although this law is nominally in force, compliance is Law 26.233 of 2007. By this time, none of the national difficult to oversee because implementing rules have programmes in the five countries restricted children’s not been adopted for it. eligibility on the basis of the mother’s employment This legislation has not changed in the period under status. However, several programmes afforded priority, consideration. Insofar as having a facility is better than in different ways, to the children of full-time working having none, the legal framework of workplace childcare mothers. This was the case of Chile’s Crece Contigo is indicative of some degree of State (if not paternal) co- (chcc) scheme, and the expansion of coverage for responsibility. However, since the services are limited to 0-3-year-olds in Uruguay. In addition, in the five countries a small group of working mothers in the formal sector, examined, a high proportion of services operated part- they are not particularly favourable for socioeconomic time.23 Clearly, these programmes’ contribution to State equity. In addition, as maternalist measures targeting co-responsibility is conditioned by the opening hours only female workers, they can disincentivize the hiring of facilities: full-day services that are compatible with of women of childbearing age, or of more women than formal working hours and that provide greater social the legal threshold above which the law obliges firms protection represent greater State co-responsibility. to provide workplace childcare facilities. Looking at the situation by country, in Argentina, The changes we see in public service provision have a reform to education legislation in 2006 established to do with extending early childhood care and education services from the age of 3 years as well as kindergartens services. Table 1 provides an analysis of the extent of for children aged between 45 days and 2 years full-time care in the framework of national programmes (Law 26.206), but these services were not implemented. for children aged 0-3 years. In each case we determine However, Child Development Centres (cedis) were eligibility criteria (fundamental for socioeconomic established for children from low-income households, equity), and the existence of coverage goals and data. provided by the State itself and by non-governmental The last two indicators serve to assess governments’ organizations (ngos) (Faur, 2011). commitment to State co-responsibility in childhood care In Brazil, the constitutional right to early education has been guaranteed since 1988, including for children

22 Law 20.744 (1974) in Argentina; article 389 of the Labour Process Code, in Decree-Law 229 of 1967 in Brazil; and article 203 of the 23 Empirical evaluation of centres and users by full-time or part-time Labour Code, in Law 19.408 in Chile. provision exceeds the scope of this analysis.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 111

TABLE 1 State co-responsibility: adoption of a full-time national programme for early childhood care and education,a between 2003 and 2013 Existing coverage, goal and hours Country Programme Eligibility criteria around 2010b

Argentina National Education Children aged 45 days or over (services 30% at age 3; 67 141 children aged 0-2, Law (2006) outside the formal education sector equivalent to 3.5% of this age group are recognized) For children aged 0-3 there were no specific measures until the creation of cedis with need-based criteria

Child Development Centres Economic need No data; target unknown (cedis) (2007)

Brazil Education Guidelines and Universal (in the framework of the In 2010, 18.4% of children were Bases Law education system) covered, with an average of 8 hours per (1996, financing since 2007) Expansion of childcare centres known day (of the target set in 2001 of 50% of as “créches” children aged 0-3, which was renewed in 2010)

Chile Chile Crece Contigo 60% most vulnerable of the population In 2010, coverage was 212 000 children, (chcc) (in the framework of care services outside meeting the target of creating 113 000 the education system) additional places Takes the form of the chcc programme

Costa Rica Day Care and Integral Need-based, aiming for universal coverage Coverage increased from 7 500 children Development Centres (in the framework of care services outside aged 0-6 in 2010 to 15 000 children in (cecudi) (2010) the education system) 2013; in keeping with target to increase cecudi scheme expanded with new and coverage by 8 000 children existing services

Uruguay National Care System (2011) Need-based aiming for universal coverage In 2009 coverage was 13% for children (can be outside the education system) under 1 year, 26% for 1-year-olds, 29% Expansion of Child and Family Assistance for 2-year-olds and 47% for 3-year-olds Centres (caif) and supply-side subsidies (a total of 41 216 places in 2008 of the target of 43 000 for 2009)

Source: Argentina: Ana Malajovich, “La exclusión de los más vulnerables: Deudas educativas con la primera infancia”, Voces en el Fénix, No. 3, Buenos Aires, University of Buenos Aires, 2014; Lea Waldmann and others, Servicios de atención a niños y niñas de 45 días a 36 meses, Buenos Aires, United Nations Children’s Fund (unicef), 2011; Brazil: National Education Plan 2001 and 2011-2020; Costa Rica: Government of Costa Rica, “Plan Nacional de Desarrollo 2010-2014”, María Teresa Obregón Zamora, San Jose, December 2010; Chile: Ministry of Planning/Ministry of Health (mideplan/minsal), Cuatro años creciendo juntos. Memoria de la Instalación del Sistema de Protección Integral a la Infancia Chile Crece Contigo 2006-2010, Santiago, Chile, 2010; Uruguay: Comité de Coordinación Estratégica de Infancia y Adolescencia, “Estrategia Nacional para la Infancia y la Adolescencia (enia) 2010-2030. Plan de Acción 2010-2015”, Working Paper, Montevideo, 2010; Ana Cerutti and others, Plan caif, 1988-2008, Montevideo, October 2008; and Soledad Salvador, “Hacia un Sistema Nacional de Cuidados en Uruguay”, Montevideo, Economic Commission for Latin America and the Caribbean (eclac), 9 December 2010. a For children aged 0-3 years. b Where hours are not indicated, the proportion correspond to full-time services is unknown. under age 1, and in 1996 this right became law. However, services. Current coverage is far from guaranteeing what has progressed during the period under analysis that sort of provision: in 2001, the National Education is the fulfilment of this legal framework: in 2006, Plan set a coverage target of 50% for children aged responsibility for early childhood care and education 0-3 for 2010. However, in 2010 coverage was less than services passed from the Ministry of Social Development 20% and the 10-year plan set the 50% target again, this and Hunger Alleviation to the Ministry of Education, time for 2020. The State performed better in terms of while implementation remained the responsibility of hours of service in childcare centres (créches); in 2012 municipalities. The main challenge since then has been daily hours averaged eight, compared with less than getting the various levels of government to enforce five in preschools (for children aged 4 and 5) (Ministry this right and, from the point of view of work-family of Education (mec)/ National Institute for Educational reconciliation, getting centres to provide full-day Studies and Research (inep)).

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 112 CEPAL REVIEW 114 • DECEMBER 2014

In 1994, Chile had a relatively limited programme centres has been complemented with public subsidies for low-income working mothers, who had to show from the Ministry of Social Development, to enable formal contracts and document their low income to children to attend existing private kindergartens. qualify for junji/Integra services (Pribble, 2006, Broadening public preschool services has taken p. 91). In 2006, the Administration of Michelle Bachelet place in parallel with changes in provision for early created the programme Chile Crece Contigo (chcc) to childhood in schools. As in the case of primary education, coordinate and broaden existing services for children this has been mainly on a part-time basis. Successive of preschool age, especially among the lower income expansions at the preschool level have occurred by quintiles. In 2009, Law 20.379 enshrined “the right to a lowering both the age at which children can start, and full-time créche and kindergarten place for the children the age at which they must start. All the countries have of mothers who are working, studying, or looking for taken steps towards universalizing access. However, the work; and the right to a part-time kindergarten place with increase in coverage has come about chiefly in part-time no further requirements regarding parents’ activities” facilities and has thus not expanded co-responsibility (Staab, 2012, p. 313) for the 60% most economically in work-family reconciliation, although it has increased vulnerable proportion of the population. co-responsibility in preschool education per se. Between 2006 and 2010, places in ecec centres Preschool education from the age of 3 has been a under the chcc umbrella more than doubled, from right in Uruguay since 1995, but was initially obligatory almost 97,000 to over 210,000 (mideplan/minsal, 2010, only from the age of 5, and later from age 4 (Pribble, pp. 59-60). With the change in government in 2010 the 2013, p. 89; Mancebo, 2012; Salvador, 2014). Costa expansion of coverage came to a halt and there were no Rica legislated on the gradual lowering of universal longer clear goals for increasing coverage. preschool education from age 5 to age 4 in 1997, In Costa Rica, the Education and Nutrition Centres/ although it was still not obligatory until age 5. In Integrated Centres for Childcare (cen-cinai) programme Argentina, attendance in preschool education from age was formalized in the 1970s and had changed little by 5 has been obligatory since 1993 (Pautassi and Zibecchi, 2010. Since then, the Government of Laura Chinchilla has 2010, pp. 18-19). In 2006, a reform legislated that free created the universal National Childcare and Development public preschool education would be made gradually Network, which brought together existing care modalities available from the age of 4 (Faur, 2008, pp. 56-57), (Sojo, 2011) and created a new municipal facility, Day although still not obligatory until age 5. In Chile, Care and Integral Development Centres (cecudi). universal preschool education has begun at age 4 since This network provides universal services for children 1997, but is not obligatory until the age of 5. In Brazil, aged 0-12, but channels public subsidies in a targeted preschool education begins at age 4 and in 2009 it was manner (imas, 2013). In 2013 the system had explicit established that attendance from that age would become eligibility criteria, as well as coverage expansion targets obligatory in 2016. —relatively modest ones, to include 8,000 children in full-time services (Government of Costa Rica, 2010)— 3. Regulatory measures: bringing the rights of including for the 0-3 age range. paid domestic workers into line with the rest In Uruguay, Child and Family Assistance Centres of the workforce (caif) have provided services to 0-4-year-olds from low-income households since 1988 (Pribble, 2006; Throughout the region, paid domestic work is a key Salvador, 2010, p. 32). Families qualify on the basis of strategy for women and families to reconcile paid social vulnerability and children’s ages (caif, 2008). work and family responsibilities (eclac, 2013b), and In 2011, the Government of José Mujica announced is therefore a useful proxy for examining the way the creation of a National Integrated Health System governments regulate occupations associated with care for children, the older adult population and persons in general. Historically, social and labour regulation has with disabilities. After solid progress in defining the legally discriminated against domestic workers. Since system’s main components, the project is currently in a most of them are women who are socioeconomically budgetary and implementation hiatus. Under this system, vulnerable, more equitable treatment of paid domestic the government plans to broaden eligibility criteria for workers is an indicator of greater socioeconomic equity. early childhood care and education services, first by In turn, since two of the rights at stake are working hours completing coverage for all children from the lowest and maternity leave, the analysis also shows progress income quintile. In the meantime, the work of the caif in State co-responsibility.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 113

In particular, it is useful to establish what reforms work, and even less, work that requires qualifications, it have been implemented to ensure that domestic workers is often thought that the regulations can dispense with have the same rights as the rest of the employed the standards and limits that exist for other occupations. population (Blofield, 2012). Figure 3 depicts the labour On the other hand, women who perform these tasks are rights of domestic workers enshrined in national labour supposed to be available unconditionally to attend to the legislation in 2003 and 2013. The comparison includes high-income families who hire them. Domestic workers’ social protection (social security and maternity leave) own family responsibilities are not taken into account in and labour regulation (minimum wage, vacation and a their condition as “servants” (Blofield, 2012). category referring to other discriminatory clauses). A value Figure 4 compares the legal working week for of 6 denotes legislation that affords domestic workers domestic workers with that of the rest of the economically the same rights as the rest of the employed population. active population in 2003 and 2013. The columns show As shown in figure 3, in 2003 the labour codes of all the differences in the hours of each occupation. In 2003, five countries discriminated against domestic workers: the difference averaged more than 20 hours in all the their working day and week were both longer than in countries.25 In 2006 in Uruguay and in 2009 in Costa other occupations. In Uruguay in 2006, in Costa Rica in Rica, these working hours were equalized. In 2013, 2009 and in Argentina and Brazil in 2013, these rights Argentina and Brazil did the same. In 2005, the gap in were brought into line with the general labour code.24 the working week of Chilean domestic workers widened As of mid-2014 —and despite a draft reform in the with respect to other workers, from 24 to 27 hours executive— only Chile maintains discriminatory clauses. (with working weeks of 72 hours in comparison with a The length of the working day has been the most reduction from 48 to 45 hours for other workers). difficult right to bring into line under the law, reflecting From the perspective of work-family reconciliation, a combination of discriminatory attitudes, both gendered a stronger guarantee of domestic workers’ labour and and socioeconomic in nature. On the one hand, domestic social rights is indicative of greater State co-responsibility and care work is seen as an activity that women perform because they are women. Since it is not considered real 25 In Brazil and Uruguay maximum working hours are not specified. The regulation suggests that these workers are expected to be available 24 These latest reforms took place after ilo adopted a convention on except for meal and rest times. Accordingly, their daily hours have the matter in 2011. been calculated at 16.

FIGURE 3 Co-responsibility and socioeconomic equity: changes in labour rights of domestic workers in labour codes, 2003 and 2013

6

5

4

3

2

1

0 Chile 2003 Chile 2013 Brazil 2003 Brazil 2013 Uruguay 2003 Uruguay 2013 Argentina 2003 Argentina 2013 Costa Rica 2003 Costa Rica 2013

Working hours No other discriminatory provision Vacation National minimum wage Obligatory and equal social security Maternity leave

Source: prepared by the authors, on the basis of Merike Blofield,Care Work and Class: Domestic Workers’ Struggle for Equal Rights in Latin America, University Park, Pennsylvania, Penn State Press, 2012, and updated as of 2013 on the basis of the national legislations of the respective countries.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 114 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 4 Difference in maximum weekly hours between domestic workers and workers in general, 2003 and 2013

60

50

40

30

20

10

0 Argentina Brazil Chile Costa Rica Uruguay

Difference in 2003 Difference in 2013

Source: prepared by the authors, on the basis of Merike Blofield,Care Work and Class: Domestic Workers’ Struggle for Equal Rights in Latin America, University Park, Pennsylvania, Penn State Press, 2012, and updated as of 2013 on the basis of the national legislations of the respective countries. in service procurement, with direct effects on work- Because guaranteeing domestic workers’ rights family reconciliation for the workers themselves. The makes this means of work-family reconciliation more socioeconomic status and gender of domestic workers, expensive for families who pay for this still highly the great majority of whom are women, are inevitably informal work, it can worsen work-family tensions for intertwined, so the way in which the State addresses their those families in the short term. Yet, in the medium labour conditions indicates the value it affords to the two and long term, formalizing domestic workers’ labour types of equity. Any measure that improves conditions for conditions can foster collective efforts towards more domestic workers promotes both dimensions of equity. institutionalized sequential and defamilializing measures.

V Analysis and conclusions

Socioeconomic inequalities have eased slightly in Latin article’s main contribution has been to break down and America in the past 10 years. However, the gap in analyse these types of reconciliation policies to show in labour participation between poor and non-poor women a simple and comparative manner how much and what changed little between the start and the end of the first type of change has occurred between 2003 and 2013 in decade of the 2000s. Because socioeconomic and gender the five countries examined, and what the implications inequalities are closely linked, reducing any type of are —from the perspective of policy design rather than inequality necessitates tackling the work-family balance policy implementation— for inequality. through public policy. Also, not all policies can promote The evidence on policy adoption shows varying socioeconomic and gender equality simultaneously. In degrees of change in the different countries and policy order to establish how much and what type of progress types. In the national comparison, in 2003 Argentina has been made, this article discussed changes in and Chile had less inclusive work-family policies, employment-based leaves, care services and the labour comparatively speaking. By 2013, Chile and particularly protection of domestic workers as indicators of sequential, Uruguay had made reforms to all three types of policy, defamilializing and regulatory policies, respectively. This whereas Argentina, at least with regard to the policies

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 115 examined and by comparison with the other countries, long in Uruguay as in Chile) and creating parental leave had made no changes. Both in 2003 and in 2013, Brazil (though with different modalities in the two countries). and Costa Rica show a mixed performance between At the same time, the Chilean reform is comparatively policy types. maternalist, insofar as its maternity leave reinforces the The comparative analysis also shows the type of idea that children are mainly a maternal responsibility. change that countries have sought to achieve. In the five In the other three countries paternity leaves have been countries measures have been taken to extend maternity on the agenda, but have not been adopted. leave to the most vulnerable groups of workers, broaden Measures aimed at guaranteeing a maternalist expectations of early childhood care and education services floor beyond maternity leave include the child benefits as a right for children, and regulate paid domestic work. established in Chile and the contribution to pensions These are very important measures for lower-income for each live birth in Uruguay, which by recognizing women and thus positive from the point of view of public the differentiated role of women aim to level their policymaking for socioeconomic equity. socioeconomic status with that of their male peers. The greater protection of paid domestic work The analysis conducted here indicates that, as a indicates to increased State co-responsibility in work- general rule, policies continue to treat the care of small family reconciliation insofar as the State acts upon the children as the responsibility of mothers. Sequential conditions in which paid care services are procured within measures (such as maternity leaves), defamilializing the household. Over the long run, from the point of view measures and the regulation of care work have all tended of the families hiring this type of labour, such measures to make more progress in promoting socioeconomic equity should also create the conditions to push the State to than in promoting paternal co-responsibility, although the design better sequential and defamilializing measures. recent reforms in Chile and Uruguay indicate a small but Early childhood education and care services qualitative change in this direction. Although the need have expanded State co-responsibility in work- for full-day early childhood care services is increasingly family reconciliation where they are provided on a on the agenda in most of the countries, performance in full-time basis. It is notable that in all five countries terms of State co-responsibility is still hard to assess. employers’ obligation to provide childcare services has The foregoing comparative analysis of change and remained unchanged. continuity has left three types of questions still to be Lastly, reforms in employment-based leave have addressed. First, how effective are these measures at the been mixed. The governments of all the countries have implementation stage? Second, what capacity do they adopted measures aimed at increasing socioeconomic have over the medium and long term to change initial equity. Two countries, Chile and Uruguay, have taken inequality and the balance between work and family? steps towards paternal co-responsibility by extending And, third, what are the social and policy determinants paternity leave (which, although still short, is twice as that drive variations between countries and policies?

Bibliography

Aguirre, Rosario and Fernanda Ferrari (2014), “La construcción del caif (Centros de Atención Integral a la Primera Infancia y la Familia) sistema de cuidados en el Uruguay: En busca de consensos para (2008), El Plan caif, Montevideo, Secretaría Ejecutiva Plan caif. una protección social más igualitaria”, Políticas Sociales series, Cerutti, Ana and others (2008), Plan caif, 1988-2008, Montevideo, No. 192 (LC/L.3805), Santiago, Chile, Economic Commission October. for Latin America and the Caribbean (eclac). Chioda, Laura (2011), Work and Family: Latin America & Caribbean Batthyany, Karina, Natalia Genta and Valentina Perrotta (2012), “La Women in Search of New Balance, Washington, D.C., World Bank. población uruguaya y el cuidado: Persistencias de un mandato Comité de Coordinación Estratégica de Infancia y Adolescencia de género. Encuesta nacional sobre representaciones sociales (2010), “Estrategia Nacional para la Infancia y la Adolescencia del cuidado: Principales resultados”, Mujer y Desarrollo series, (enia) 2010-2030. Plan de Acción 2010-2015”, Documento de No. 117 (LC/L.3530), Santiago, Chile, Economic Commission Trabajo, Montevideo. for Latin America and the Caribbean (eclac). Cornia, Giovanni (2010), “Income distribution under Latin America’s Bentancor, Andrea and María Isabel de Martini (2012), Detrás de la new left regimes”, Journal of Human Development and puerta: Trabajo, roles de género y cuidado, Santiago, Chile, Capabilities, vol. 11, No. 1, Taylor & Francis, February. Comunidad Mujer. Durán, María-Ángeles (2004), “¿Cómo conciliar trabajo y vida Blofield, Merike (2012), Care Work and Class: Domestic Workers’ familiar? Un desafío colosal” [online] http://www.comfia.info/ Struggle for Equal Rights in Latin America, University Park, index.php?modo=leer&art=14814. Pennsylvania, Penn State Press. eclac (Economic Commission for Latin America and the Blofield, Merike and Juliana Martínez Franzoni (2014), “Maternalism, Caribbean) (2013a), Social Panorama of Latin America, 2013 co-responsibility and social equity: a typology of work-family (LC/G.2580), Santiago, Chile. United Nations publication, Sales policies”, Social Politics, June. No. E.14.II.G.6.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. 116 CEPAL REVIEW 114 • DECEMBER 2014

(2013b), Social Panorama of Latin America, 2012 imas (Joint Institute for Social Aid) (2013), Monto de los subsidios (LC/G.2557-P), Santiago, Chile. United Nations publication, para la participación en las diferentes alternativas de cuido Sales No. E.13.II.G.6. y desarrollo infantil y directrices técnicas de cooperación, (2011), “Childcare and parental leave”, Challenges: seps-022-1-13, San Jose, January. Newsletter on Progress Towards the Millenium Development Koven, Seth and Sonya Michel (1993), Mothers of a New World: Goals from a Child Rights Perspective, No. 12, Santiago, Chile, Maternalist Politics and the Origins of Welfare States, New Economic Commission for Latin America and the Caribbean York, Routledge. (eclac)/United Nations Children’s Fund (unicef), July. López-Calva, Luis and Nora Lustig (2010), Declining Inequality (2010), Social Panorama of Latin America, 2009 in Latin America: a Decade of Progress?, Washington, D.C., (LC/ G.2423-P), Santiago, Chile. United Nations publication, Brookings Institution. Sales No. E.09.II.G.135. Lupica, Carina (2013), Trabajo decente y cuidado compartido: Hacia England, Paula and Nancy Folbre (1999), “The cost of caring”, una propuesta de parentalidad, Santiago, Chile, International The Annals of the American Academy of Political and Social Labour Organization (ilo)/United Nations Development Science, vol. 561, No. 1, American Academy of Political and Programme (undp). Social Science. Malajovich, Ana (2014), “La exclusión de los más vulnerables: Esping-Andersen, Gøsta (2009), The Incomplete Revolution: Adapting Deudas educativas con la primera infancia”, Voces en el Fénix, to Women’s New Roles, Cambridge, The Polity Press. No. 3, Buenos Aires, University of Buenos Aires. (1990), The Three Worlds of Welfare Capitalism, Cambridge, Mancebo, Esther (2012), Personal Communication, 13 July. Polity Press. Martínez Franzoni, Juliana and Rosalía Camacho (2007), Faur, Eleonor (2011), “A widening gap? The political and social “Equilibristas o malabaristas..., pero ¿con red? La actual organization of childcare in Argentina”, Development and infraestructura de cuidados en América Latina”, Entre familia y Change, vol. 42, No. 4, Wiley. trabajo: Relaciones, conflictos y políticas de género en Europa (2008), The “Care Diamond”: Social Policy Regime, Care y América Latina, María Antonieta Carbonero and Silvia Levín Policies and Programmes in Argentina, Geneva, United Nations (eds.), Rosario, Homo Sapiens Ediciones. Research Institute for Social Development (unrisd). (2006), “Género, políticas conciliatorias y presupuesto Faur, Eleonor, Valeria Esquivel and Elizabeth Jelin (eds.) (2012), público: Una aproximación al caso de América Latina y el Las lógicas del cuidado infantil: Entre las familias, el Estado Caribe”, Cohesión social, políticas conciliatorias y presupuesto y el mercado, Buenos Aires, Institute of Economic and público. Una mirada desde el género, Luis Mora, María José Social Development. Moreno and Tania Rohrer (coords.), Mexico City, United Filgueira, Fernando (2011), “Fault lines in Latin American social Nations Population Fund (unfpa)/German Agency for development and welfare regime challenges”, The Great Technical Cooperation (gtz). Gap: Inequality and the Politics of Redistribution in Latin Martínez Franzoni, Juliana (2008), “Welfare regimes in Latin America, Merike Blofield (ed.), College Park, Pennsylvania, America: Capturing constellations of markets, families and Pennsylvania State University Press. policies”, Latin American Politics and Society, vol. 50, Folbre, Nancy (1995), “Holding hands at midnight: the paradox of No. 2, Wiley, June. caring labor”, Feminist Economics, vol. 1, No. 1, Taylor & Martínez Franzoni, Juliana and Diego Sánchez-Ancochea (2013), Francis, February. “Public social goods and income inequality”, A Primer Fraser, Nancy (1997), Justice Interruptus: Critical Reflections on the on Labour Market Institutions and Inequality, Geneva, “Postsocialist” Condition, London, Routledge. International Labour Organization (ilo), in press. Gherardi, Natalia, Laura Pautassi and Carla Zibecchi (2012), De mideplan/minsal (Ministry of Planning/Ministry of Health) (2010), eso no se habla: El cuidado en la agenda pública. Estudio Cuatro años creciendo juntos. Memoria de la Instalación del de opinión sobre la organización del cuidado, Buenos Aires, Sistema de Protección Integral a la Infancia Chile Crece Equipo Latinoamericano de Justicia y Género (ela). Contigo 2006-2010, Santiago, Chile. Goldin, Claudia (2006), “The quiet revolution that transformed Ministry of Education (2011), “Plano Nacional da Educaçao women’s employment, education, and family”, American 2011-2020”, Brasilia. Economic Review, vol. 96, No. 2, Nashville, Tennessee, (2001), “Plano Nacional da Educaçao 2001”, Brasilia. American Economic Association, January. Ministry of Social Development (2011), Plan Nacional de Acción Government of Chile (2009), “Sistema de protección a la infancia: por los Derechos de Niños, Niñas y Adolescentes 2012-2015, Chile Crece Contigo”, Santiago, Chile. Buenos Aires. Government of Costa Rica (2010), “Plan Nacional de Desarrollo Monge, Guillermo (2006), “Igualdad de género, pobreza y políticas 2010-2014”, María Teresa Obregón Zamora, San Jose, de conciliación entre los ámbitos productivo y reproductivo y December. presupuestos públicos”, Mexico City. Hirata, Helena and Nadya Araujo Guimarães (eds.) (2012), Cuidado Montaño Virreira, Sonia (2010), “El cuidado en acción”, El cuidado e cuidadoras: As várias faces do trabalho do care, São Paulo, en acción: Entre el derecho y el trabajo, Cuadernos de la Editora Atlas. cepal series, No. 94 (LC/G.2454-P), Sonia Montaño and Huber, Evelyne and John Stephens (2012), Democracy and the Left: Coral Calderón Magaña (eds.), Santiago, Chile, Economic Social Policy and Inequality in Latin America, Chicago, The Commission for Latin America and the Caribbean (eclac). Chicago University Press. United Nations publication, Sales No. S.10.II.G.35. ilo (International Labour Organization) (2012), Panorama Laboral Orloff, Ann Shola (2009), “Gendering the comparative analysis of 2012. América Latina y el Caribe, Lima, ilo Regional Office welfare states: an unfinished agenda”, Sociological Theory, for Latin America and the Caribbean. vol. 27, No. 3, Wiley, September. (2011), Panorama Laboral 2010. América Latina y el (2006), “Farewell to maternalism? State policies and Caribe, Lima, ilo Regional Office for Latin America and mothers’ employment”, The State After Statism, Jonah Levy the Caribbean. (ed.), Cambridge, Massachusetts, Harvard University Press. ilo/undp (International Labour Organization/United Nations Pautassi, Laura and Carla Zibecchi (2010), “La provisión de Develoment Programme) (2009), Trabajo y familia: Hacia cuidado y la superación de la pobreza infantil. Programas de nuevas formas de conciliación con corresponsibilidad social, transferencias condicionadas en Argentina y el papel de las Santiago, Chile. organizaciones sociales y comunitarias”, Políticas Sociales

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F. CEPAL REVIEW 114 • DECEMBER 2014 117

series, No. 159 (LC/L.3198-P), Santiago, Chile, Economic (2013), “Análisis de costos y posibles impactos de Commission for Latin America and the Caribbean (eclac). diferentes modelos de licencias por maternidad, United Nations publication, Sales No. S.10.II.G.10. paternidad y parentales. Informe final de consultoría en el Pautassi, Laura and Nieves Rico (2011), “Childcare leave: a right marco del proyecto URY2U503”, Documento de Trabajo, of children and parents”, Challenges: Newsletter on Progress 15 February. Towards the Millenium Development Goals from a Child Rights (2010), “Hacia un Sistema Nacional de Cuidados en Perspective, No. 12, Santiago, Chile, Economic Commission Uruguay”, Montevideo, Economic Commission for Latin for Latin America and the Caribbean (eclac)/United Nations America and the Caribbean (eclac), 9 December. Children’s Fund (unicef), July. Sauma, Pablo (2012), “Protección social y trabajo no remunerado: pnad/ibge (National Household Survey/Brazilian Geographical Redistribución de las responsabilidades y tareas del cuidado. and Statistical Institute) (2014) [online] http://www. ibge. Estudio de caso Costa Rica”, Mujer y Desarrollo series, gov.br/home/estatistica/indicadores/trabalhoerendimento/ No. 116 (LC/L.3519), Santiago, Chile, Economic Commission pnad_continua/default.shtm. for Latin America and the Caribbean (eclac). Presidency of the Federative Republic of Brazil (2009), “Emenda Sojo, Ana (2011), “De la evanescencia a la mira: El cuidado como constitucional 59”, 11 November. eje de políticas y de actores en América Latina”, Seminarios Pribble, Jennifer (2013), Between Elites and the Masses: Political y Conferencias series, No. 67 (LC/L.3393), Santiago, Parties and Welfare Regimes in Latin America, Cambridge, Chile, Economic Commission for Latin America and the Cambridge University Press. Caribbean (eclac). (2006), “The politics of women’s welfare in Chile and Sorj, Bila (2013), “Arenas de cuidado nas interseções entre género Uruguay”, Latin American Research Review, vol. 41, No. 2, e classe social no Brasil”, Cadernos de Pesquisa, vol. 43, Latin American Studies Association (lasa), December. No. 149, São Paulo, May-August. Ray, Rebecca, Janet Gornick and John Schmitt (2010), “Who cares? Sorj, Bila and Adriana Fontes (2012), “O care como um regime Assessing generosity and gender equality in parental leave estratificado: Implicações de género e classe social”,Cuidado policy designs in 21 countries”, Journal of European Social e cuidadoras. As várias faces do trabalho do care, Helena Policy, vol. 20, No. 3, Sage. Hirata and Nadya Araujo Guimaraes (orgs.), São Paulo, Atlas. Rodríguez, Corina and Laura Pautassi (coords.) (2014), La Staab, Silke (2012), “Maternalism, male-breadwinner bias and market organización social del cuidado de niños y niñas: Elementos reform: historical legacies and current reforms in Chilean social para la construcción de una agenda de cuidados en policy”, Social Politics, vol. 19, No. 3. Argentina, Buenos Aires, Association for Civil Rights (adc)/ Waldmann, Lea and others (2011), Servicios de atención a niños y Interdisciplinary Centre for Public Research (ciepp)/Equipo niñas de 45 días a 36 meses, Buenos Aires, United Nations Latinoamericano de Justicia y Género (ela). Children’s Fund (unicef). Román, Isabel and Natalia Morales (2010), “Demanda potencial Williams, Fiona (2010), “Claiming and framing in the making of de cuido infantil según estructura de los hogares: El caso de care policies: the recognition and redistribution of care”, Costa Rica”, San Jose. Gender and Development Programme, No. 13, Geneva, Salvador, Soledad (2014), “Personal communication”, Montevideo, United Nations Research Institute for Social Development 4 April. (unrisd), November.

WORK, FAMILY AND PUBLIC POLICY CHANGES IN LATIN AMERICA: EQUITY, MATERNALISM AND CO-RESPONSIBILITY • MERIKE BLOFIELD AND JULIANA MARTÍNEZ F.

A first approach to the impact of the real exchange rate on industrial sectors in Colombia

Lya Paola Sierra and Karina Manrique L.

ABSTRACT Much has been said about possible symptoms of Dutch disease in Colombia in the wake of a marked upsurge in commodity prices and the significant real appreciation of the national currency. This paper examines whether the real effective exchange rate had an impact on industry during the period 2000-2010. Specifically, it evaluates the effect of the appreciation of the real exchange rate on the value added of 63 industrial sectors

in Colombia using the Arellano and Bond (1991) generalized method of moments (gmm) estimator. Overall, our results confirm a negative relationship between real exchange rate appreciation and industry. The analysis showed that real exchange rate appreciation had a significant impact on the value added of 21 sectors: a negative effect for 18 sectors and a positive effect for 3 sectors.

KEYWORDS Currency instability, industry, industrial development, competitiveness, econometric models, statistical data, Colombia

JEL CLASSIFICATION O24, L60, F31

AUTHORS Lya Paola Sierra is an assistant professor at the Department of Economics of the Pontifical Javeriana University in Cali, Colombia. [email protected] Karina Manrique L. is a Professor with the industrial engineering curriculum project at the Francisco José de Caldas District University in Bogota, Colombia. [email protected] 120 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

In 2003, the Colombian currency began one of the most this problem, newspapers and journals, encouraged marked periods of real appreciation in the country’s recent by the industrial sector, have recently issued warnings history. Apart from the decrease recorded during one year about the possible contagion of the Dutch disease in owing to the global financial crisis, the real exchange Colombia. The possible effects on manufacturing of rate appreciated by 51% in the period 2003-2011. This this real appreciation of the exchange rate in Colombia is the seventh highest rate of real appreciation out of a have motivated this research, which sets out to determine group of 95 countries, according to a World Bank index the effect of the real exchange rate on industrial value of real exchange rates. added for the period 2000-2010. One factor that contributed to the appreciation was We used data for 63 sectors from the Annual the dramatic increase in oil prices during the period. Since Manufacturing Survey, along with macroeconomic oil makes up about half of the country’s total exports, the data, to conduct estimations using the Arellano and 275% rise in real oil prices resulted in windfall profits Bond (1991) generalized method of moments (gmm) and pushed up the nominal exchange rate. estimator. We used cross-departmental, cross-sectoral The real exchange rate can influence the information to evaluate the impact of real appreciation competitiveness of industrial products in international on each of the industrial sectors in Colombia. markets. As the Colombian peso appreciates in real terms The rest of the paper is organized as follows: in the it drives up the prices of local goods with respect to those next section, we briefly review the relevant literature; from the rest of the world. This makes Colombian products in section III we review the data and the econometric less competitive than their overseas competitors and approach used in the study; in section IV we present the may have a negative impact on output and employment results of the model; and lastly, conclusions are drawn in sectors that produce tradable goods. In response to in section V.

II Literature review

The real appreciation of the peso against the dollar since keynote speech given at the headquarters of the Economic 2003 (see figure 1) has raised fears among politicians and Commission for Latin America and the Caribbean (eclac) the industrial sector of de-industrialization in Colombia. in Chile: As the period of appreciation coincided with the great “We are trying to attract investors to sectors other than upsurge in real oil prices (see figure 2) —a major oil and mining because we are now facing a prelude export for Colombia— national newspapers started to to the Dutch disease owing to the concentration of debate whether Colombia was already suffering from investment in these sectors.” symptoms of Dutch disease.1 Many economic analysts President Juan Manuel Santos, 17 August 2011. have written columns on this subject and the President However, despite the media interest in the subject, of Colombia even made a reference to the topic in a few articles have been written on the possible symptoms of Dutch disease caused by the recent real appreciation of the currency in Colombia. Lya Paola Sierra gratefully acknowledges the financial support Dutch disease refers to the fallout caused by windfall received from the Pontifical Javeriana University in Cali. profits from a resource discovery (Corden and Neary, 1 Other possible causes of the real appreciation of the exchange rate 1982; Corden, 1984; Beverelli, Dell’Erba and Rocha, could have included the positive trends in foreign direct investment, 2011), a resource price boom (Egert and Leonard, 2008; the increased privatization of State agencies and, externally, excess liquidity in the United States and Europe. Algieri, 2011; Poncela, Senra and Sierra, 2012), an

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 121

FIGURE 1 Colombia: real effective exchange rate index, 1990-2011 (Index 2005 = 100)

130

120

110

100

90

80

70

60 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Source: World Bank.

FIGURE 2 Annual average Brent crude oil spot prices, 1990-2010 (Dollars per barrel)

100

90

80

70

60

50

40

30

20

10

0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Source: Energy Information Administration. upsurge in remittances (Acosta, Lartey and Mandelman, The Economist in 1977, which described the detrimental 2009; Guha, 2013) or higher capital inflows in the form impact on the industrial sector in the Netherlands following of foreign aid or foreign direct investment (Lartey, 2011; the discovery of major gas deposits in the North Sea. Arellano and others, 2005; Prati and Tressel, 2005).2 Corden and Neary (1982) formulated the core theoretical The term “Dutch disease” is attributed to an article in model of Dutch disease. They used a Salter-Swan framework (Salter, 1959; Swan, 1960) to describe how windfall profits in a country can cause real appreciation 2 Buiter and Purvis (1983) examine the relative importance of different followed by reductions in competitiveness and output in shocks as causes of de-industrialization, by looking at the impact on the non-resource tradable sector. A resource boom may the real exchange rate of factors including higher oil prices, a domestic lead to a real appreciation through two channels. First, oil discovery and monetary disinflation.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 122 CEPAL REVIEW 114 • DECEMBER 2014 newfound wealth can lead to higher rates of national output could be maintained through the higher domestic absorption, through spending either by the government demand associated with the resource boom. That is, or directly by the owners of the factors. This increase losses in competitiveness would be compensated by in demand drives up the prices of non-tradable goods, gains in domestic demand. forcing real appreciation. Second, the nominal exchange Moreover, real appreciation can potentially increase rate can appreciate in a flexible exchange rate regime competitive pressures and force industrial restructuring, owing to burgeoning inflows to the country. Given the which can in turn boost productivity. The effect of the traditional real exchange rate equation (see equation real exchange rate on manufacturing output depends (1)), where the real exchange rate (Q) is represented in on the exposure of industries to international markets. the equation as the price of domestic goods (P) relative Industries that export most of their output, for example, to those from abroad (P*) adjusted by the nominal might see their profits fall as they lose competitiveness exchange rate (S), both forces lead to real exchange rate in periods of real appreciation. Whereas industries that appreciation. Corden and Neary (1982) referred to this import most of their intermediate inputs could benefit as the “spending effect”. from local currency appreciation, since real appreciation tends to make these inputs cheaper. To summarize, a P real appreciation shock has an ambiguous effect on QS= * (1) P* profitability and industrial performance. Some of the papers that examine the real exchange rate and industrial performance include Burgess and Knetter (1998); Campa Taking logs, the real exchange rate is also denoted as: and Goldberg (1995 and 2001); Goldberg, Tracy and Aaronson (1999); Goldberg (1993); Campbell and * qstt=+pptt− (2) Lapham (2004); Ekholm, Moxnes and Ulltveit-Moe (2012), and Berman, Martin and Mayer (2012). Here, st is the log of the foreign currency price of Although the relationship between the real exchange domestic currency (United States dollars per Colombian rate and industrial output remains unclear, there appears * peso), pt and pt are the logs of the national and foreign to be strongly supported cross-country statistical evidence country price indices, respectively. In this definition of that overvalued currencies are associated with slow the real exchange rate, an increase in qt means a real growth, especially in less developed countries (Rajan appreciation of the local currency and a decrease in qt and Subramanian, 2011; Rodrik, 2008, and Berg, Ostry reflects a real depreciation. and Zettelmeyer, 2012). Tradable sectors, particularly Apart from the “spending effect”, Corden and those in manufacturing, seem to be the link between Neary (1982) describe a “resource movement effect”, the real exchange rate and economic growth. Rodrik referring to the reallocation of factors, especially labour, (2008), for example, shows that the bigger a tradable from the manufacturing sector to the booming resource industrial sector is in a less developed country, the more sector. The reduction of labour in the manufacturing overvaluation hurts growth. sector contributes to the de-industrialization process. In the case of Colombia, there are few articles The literature seems to point to a clear causal link that examine the real exchange rate in relation to between a resource boom and a real appreciation of the manufacturing performance. The article bearing the national currency. However, the subsequent link between closest resemblance to this study is Echavarría and real appreciation and relative de-industrialization remains Arbeláez (2003), who measured the effect of the unclear (see Magud and Sosa (2010) for a judicious exchange rate on investment, sales and profits in review of the literature on Dutch disease). In fact, a Colombian companies in 1994-2002. Unlike our article, theoretical article by Buiter and Purvis (1983) posits however, Echavarría and Arbeláez (2003) took into that a resource boom might have a positive effect on account a devaluation period in a firm-level analysis, manufacturing on the basis of the coexistence of real which included only manufacturing firms. Carranza appreciation and an upsurge in growth. Manufacturing and Moreno (2013) analysed the vertical industrial is thus contemporaneously influenced by appreciation, chain of Colombia for the period 1990-2010, evaluating which shrinks international competitiveness, and local possible industrialization, but did not specifically demand for manufacturing, which pushes up sales. Since assess the effect of the exchange rate on industry. Other Colombia, which is a small economy, takes the world studies addressing the subject include Clavijo (1990), price of manufactures as a constant, manufacturing which evaluated the effect of the real exchange rate on

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 123 productivity, and Rhenals and Saldarriaga (2007), which real exchange rate on manufacturing value added. explored the relationship between the real exchange Further research should be carried out to evaluate the rate and Colombia’s economic growth. In this paper we potential impact of a resource boom on economic growth concentrate exclusively on the effects of the Colombian in Colombia.

III Econometric approach and data

1. Data products contracted from 7.5% to 7.2%. By contrast, the manufacture of furniture and the manufacture of coke We used annual data from the Annual Manufacturing and refined petroleum products increased their share in Survey conducted by the National Administrative total value added between 2000 and 2010. The furniture Department of Statistics (dane) of Colombia. The sector accounted for 7.7% of total value added in 2000, survey contains information on 63 industrial sectors in increasing to 8.7% in 2010, and the manufacture of coke 23 departments (geographical areas) in Colombia from and refined petroleum products saw the largest increase 2000 to 2010. The names of the sectors are listed in annex in the whole sample of sectors, from 7.9% to 12.8%. 1. We used variables from the Annual Manufacturing As we can see from this information, industrial Survey such as employment, number of firms per industry value added is concentrated in a small number of sectors and wages paid to personnel. in Colombia. The top five sectors accounted for 67% In addition, we used data on macroeconomic of total value added in 2000, and by 2010 their share variables such as the real effective exchange rate from had risen to 70%. the International Monetary Fund, the per capita income for each department from dane and the lending rate 2. The model from the central bank of Colombia. We constructed the ratio of imported intermediate goods, which takes into We propose the following model: account the ratio of industrial intermediate imports to total intermediate goods, with a view to controlling for yw=+bb ++bbiqI++bfIR (3) each industry’s openness to foreign markets.3 dstd01st 23tt4 dstit As a first look at the evolution of the sectors from 2000 to 2010, we grouped the data using the two-digit yqdstt=+ab02Drsd++bb32gdptt i numerical codes of the International Standard Industrial (4) Classification of All Economic Activities isic( ) adapted ++bb46wIdstdIR st+ f it for Colombia by dane (see annex 2). According to this 23 DD= information, the sector with the greatest value added where si/1 . in 2000 was the manufacture of food products and The variable y represents the value added of beverages, followed by the manufacture of chemicals industrial sector s in department d, in year t. The real and chemical products; the manufacture of coke, refined exchange rate is represented by the variable qt and Ds petroleum products and nuclear fuel; the manufacture is a dummy per industry. The rest of the variables are: of furniture; and the manufacture of other non-metallic real per capita income for each department, rgdpdt, mineral products. From 2000 to 2010, the top two sectors, real wage per industry wdst, real interest rate, it, and the food and chemicals, saw their share in the total value variable IIRdst representing the intermediate input ratio. added slide from 28% to 27% and from 16% to 14%, In equation (3) we aim to measure the overall impact respectively. Likewise, the contribution to total value of the real exchange rate on the industry in general. added of the manufacture of other non-metallic mineral The marginal effect of fluctuations in the real effective exchange rate, whether real appreciation or depreciation, on each industry is captured by parameter b2 in 3 All data have been log-transformed. equation (4).

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 124 CEPAL REVIEW 114 • DECEMBER 2014

We are aware of some known identification problems In equation (5), as in equations (3) and (4), lags regarding the estimation, such as the multi-causality of of the dependent variable are taken as explanatory industrial value added and the real exchange rate, as well variables. This fact introduces bias to the estimation by as the possibility of multicollinearity between dependent ordinary least squares (ols), since these violate the strict variables. In order to address these problems, we estimated exogeneity assumption. To tackle this issue Anderson and a dynamic linear panel data model, using the Arellano- Hsiao (1981) and Arellano and Bond (1991) suggested Bond gmm estimator. The advantage of this model is that differencing the model and then using instrumental it relaxes the strong exogeneity assumption, allowing variables estimations. the explanatory variables to be correlated with the error term. The strategy is to use the lags of the variables as DDYY=+abDDXv' + (6) instruments. A brief analysis of the characteristics and it it, - 1 it it assumptions of this model is shown below. Consider a model that includes the lag of the By transforming the regressors by first-differencing, dependent variable, Yit as a regressor (the dynamics as shown in equation (6), the fixed specific effect,η i, is introduced in the model are given by this feature). The removed, because it does not vary with time. We follow basic dynamic autoregressive model panel data can be Arellano and Bond (1991) and use the gmm estimator, represented as follows: which takes into account the passing information from Y and X as instruments. As a robustness check, we also estimated the YYit =+abit, - 1 Xu'it ++hiit (5) Arellano and Bover (1995) system gmm estimator. According to these authors, if the autoregressive process where t = 1,…,T. is persistent, or when T (number of years) is small, then 'X it is the row vector of observed explanatory the lagged levels are weak instruments. They proposed variables for individual i at time t; b is the vector of using additional moment conditions in which lagged parameters to be estimated; ηi represents the time invariant differences of the dependent variable are orthogonal to individual effect and uit, represents the idiosyncratic errors. levels of the disturbances.

IV Estimation results

Before looking at the results disaggregated by industrial for departmental per capita income, whereas those in sector, we first show the results for equation (3), which column (1) are not. The lower part of table A.3.1 includes give an initial overview of the effect of the real exchange the results of the Sargan test and Arellano-Bond test rate on industry.4 used to evaluate the overidentifying conditions and the We obtained estimates of equation (3). We report serial correlation in the first-differenced disturbances. the results for the two-step gmm estimator for both the The results show that fluctuations in the real first-differenced equation and the system equation. exchange rate significantly affect the industrial sector in We take as instruments the lagged levels dated t-2 and general. In fact, a 1% appreciation of the real exchange earlier. As additional instruments, we take the lagged rate produces a 0.29% decrease in value added, ceteris differences dated t-1. The estimation results are reported in paribus, in both the first-differences and systemgmm annex 3. Annex table A.3.1 provides estimates for equation estimations. The real appreciation of the Colombian peso (3) using the first-differences gmm and the system gmm creates a loss of competiveness in international markets estimator. The results in column (2) are controlled since local prices are higher than those of international competitors. Also, domestic consumers replace expensive national goods with cheaper imports. 4 In this section we refer to the sectors listed in table A.1.1 of annex When departmental per capita income is controlled 1, disaggregated at the three-digit level of the International Standard Industrial Classification of All Economic Activities (isic), Rev. 3, for, the results of the estimations for equation (3) adapted for Colombia by dane. show that the real exchange rate has slightly less of an

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 125 impact on the value added of the industrial sector. An In sum, the 18 sectors that are hit by appreciation appreciation of 1% generates a reduction in the value account for approximately 53% of total value added, on added of 0.26% in both the first-differences and system average, between 2000 and 2010. Conversely, the three gmm estimations (columns numbered (2) in table A.3.1). sectors that benefit from appreciation represent 4% of Interestingly, even though there is a negative effect of an total value added (see annex table A.3.4). Our results appreciation in the same year, we found that industrial show, therefore, that real appreciation is detrimental to the value added is affected significantly and in a positive sectors representing more than half of total manufacturing way by the lag of the real exchange rate. value added; however, for the 38 sectors that account With respect to the results disaggregated by for 44.8% of total manufacturing value added, the real industrial sector corresponding to equation (4), as with exchange rate does not have a significant effect. This previous estimations, we report results for the two-step non-significance may have to do with the degree of gmm estimator for the first-differenced equation, using openness of the Colombian economy: from 2000 to as instruments the lagged levels dated t-2 and earlier. 2010, trade accounted for a 35% share in total gross We evaluated two models: the first (in column (1) in domestic product (gdp) on average. An appreciation table A.3.2) does not take into account the variable of of the real exchange rate erodes the competitiveness of departmental per capita income, while the second model domestic firms in the international market, which reduces 5 (in column (2)) does control for the rgdpt variable. net exports and shifts part of domestic demand from We found that the real effective exchange rate domestic goods to foreign goods. As a consequence, had significant marginal effects in 21 industries. Real according to the Dutch disease hypothesis, this leads to appreciation generated a reduction in the value added of a drop in production and employment. However, when 18 of those industries. The manufacturing sectors that were the degree of openness of the economy is not large, as hit hardest were: television and radio receivers, veneer is the case in Colombia, these effects do not necessarily sheets, finishing of textiles not produced in the same spread to all the manufacturing sectors. Moreover, for production unit, rubber products, non-metallic mineral manufacturing, the domestic market is more important products, and sugar mills and refineries. In contrast, than the external market. For the period 2000-2009, appreciation generated positive effects in the value added of for example, domestic sales represented about 83% of only three sectors: manufacture of optical instruments and total manufacturing sales, according to dane. A further photographic equipment, publishing and the manufacture avenue for research would be to evaluate the long-term of insulated wire and cable. Annex table A.3.3 contains impact of the real exchange rate on manufacturing as a summary of the sectors that are significantly affected the country becomes increasingly open to trade. This by fluctuations in the real exchange rate (only the sectors paper should persuade policymakers to consider the significantly affected by the variableq t in equation (4)). possible impacts of real exchange rate fluctuations on It is striking that most of the sectors that are manufacturing in an economy that is wide open to trade. negatively influenced by the real exchange rate are those It is important to note that the share of manufacturing that account for the largest share in industrial value added. in gdp for the period 2000-2010 was only 15.4% on The exception is the manufacture of furniture, which is not average, while services accounted for 59.3%. Therefore, significantly affected by real exchange rate fluctuations. we cannot be conclusive about the effects of the real Surprisingly, the value added of the manufacture of exchange rate on Colombia’s gdp. In fact, as shown in refined petroleum products has a negative relationship figure 3, there was a huge decline in the contribution with qt, that is, a real appreciation of 1% leads to a loss of industry to Colombian gdp in the 1990s, coinciding in the value added of this sector of about 1.13%. with the introduction of trade liberalization policies. The specification tests do not produce evidence In our period of analysis, from 2000 to 2010, industry against any model. The Sargan test leads to non-rejection recovered to some extent. According to some authors, of the null hypothesis of that model and overidentifying such as Echavarría and Villamizar (2006), the de- conditions are correctly specified. Furthermore, the industrialization process in Colombia started in 1960, autocorrelation tests c1 and c2 (see notes to tables with the decrease in industry’s share of employment, A.3.1 and A.3.2) are consistent with the structure that and in 1970, with the decrease in its share of overall we proposed for the idiosyncratic error term. production. Echavarría and Villamizar (2006), as well as Poncela, Senra and Sierra (2012), did not find evidence of de-industrialization related to Dutch disease in the 5 We also estimated the system gmm model in this equation, however, long term. the Sargan statistic rejected the different models we proposed.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 126 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 3 Colombia: share of services and industry in gross domestic product, 1965-2011 (Percentages)

65

60

55

50

45

40 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011

Services Industry Source: Data Service & Information, [online] http://www.dsidata.com/default/page/slug/about. V Conclusions

In view of the national interest in the possible impact most markedly were those with a larger share in total on industry of real appreciation caused by the upsurge industrial value added. Overall, the sectors that suffered in oil prices after the year 2000, we sought to evaluate as a result of the real appreciation of the Colombian peso one of the symptoms of the Dutch disease by analysing accounted for 53% of total manufacturing value added, the impact of the appreciation of the real effective 39% of all manufacturing employees and 36% of firms exchange rate on the value added of 63 industrial sectors in the manufacturing sectors. in Colombia during the period 2000-2010. The results in this paper provide initial insight into We used the annual panel data set of the Annual the effects of the real exchange rate on the industrial Manufacturing Survey carried out by dane, and conducted sectors in Colombia. Although this article cannot confirm estimations using the first-differenced gmm estimator of an acceleration of the de-industrialization process in the Arellano and Bond (1991). period of study because a large number of sectors were Our results suggest that the real exchange rate had not affected by the real exchange rate, it gives a list of a significant impact on the industrial sector in general. sectors that are potentially sensitive to fluctuations in We found that the effect was negative: that is, a 1% the real exchange rate. Government policy should focus appreciation of the real exchange rate produced a decrease special attention on these sectors, which are potentially in industrial value added of between 0.26% and 0.29%. harmed in periods of appreciation. Measures such as tax The estimation results for individual industrial breaks or credit facilities could provide temporary relief sectors showed that real exchange rate appreciation had to those sectors. Further research is needed to examine a negative impact on 18 sectors and a positive impact on how the real exchange rate affects other important 3 sectors. It is striking that the real exchange rate had variables such as industrial employment, productivity no significant impact on most of the sectors (38) during and number of firms per sector in order to evaluate the the period 2000-2010. The sectors that were affected total impact of the real exchange rate on manufacturing.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 127

ANNEX 1

Table A.1.1

Sectors included in the study

International Standard Industrial Classification of Sector All Economic Activities (isic) Rev. 3 adapted for Colombia

Production, processing and preservation of meat and fish 151 Processing of fruits, vegetables, oils and fats 152 Manufacture of dairy products 153 Production of grain mill products, starches and starch products, and prepared animal feeds 154 Manufacture of bakery products, macaroni, noodles, couscous and similar farinaceous products 155 Manufacture of coffee 156 Sugar mills and refineries 157 Manufacture of other food products 158 Manufacture of beverages 159 Manufacture of tobacco products 160 Preparation and spinning of textile fibres 171 Weaving of textiles 172 Finishing of textiles not produced in the same production unit 173 Manufacture of other textile products 174 Manufacture of knitted and crocheted fabrics and articles 175 Manufacture of wearing apparel, except fur apparel 181 Tanning and preparation of leather 191 Manufacture of footwear 192 Manufacture of travel goods, handbags and similar articles 193 Sawing, planing and impregnation of wood 201 Manufacture of veneer sheets, manufacture of plywood, laminboard, particle board 202 and other panels and boards Manufacture of builders’ carpentry and joinery 203 Manufacture of wooden containers 204 Manufacture of other products of wood, manufacture of articles of cork, straw and plaiting materials 209 Manufacture of paper, cardboard and paper and cardboard products 210 Publishing 221 Printing 222 Service activities related to printing 223 Manufacture of coke oven products 231 Manufacture of refined petroleum products 232 Manufacture of basic chemicals 241 Manufacture of other chemical products 242 Manufacture of synthetic and artificial fibres 243 Manufacture of rubber products 251 Manufacture of plastics products 252 Manufacture of glass and glass products 261 Manufacture of non-metallic mineral products n.e.c. 269 Manufacture of basic iron and steel 271 Manufacture of basic precious and non-ferrous metals 272 Manufacture of structural metal products, tanks, reservoirs and steam generators 281 Manufacture of other fabricated metal products and related metalworking service activities 289 Manufacture of general purpose machinery 291 Manufacture of special purpose machinery 292 Manufacture of domestic appliances n.e.c. 293 Manufacture of office, accounting and computing machinery 300 Manufacture of electric motors, generators and transformers 311 Manufacture of electricity distribution and control apparatus 312

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 128 CEPAL REVIEW 114 • DECEMBER 2014

Table A.1.1 (conclusion)

International Standard Industrial Classification of Sector All Economic Activities (isic) Rev. 3 adapted for Colombia Manufacture of insulated wire and cable 313 Manufacture of accumulators and electrical batteries 314 Manufacture of electric lamps and lighting equipment 315 Manufacture of other electrical equipment n.e.c. 319 Manufacture of electronic valves and tubes and other electronic components 321 Manufacture of television and radio receivers, sound or image recording or reproducing 323 apparatus, and associated goods Manufacture of medical appliances and instruments and appliances for measuring, 331 checking, testing, navigating and other purposes, except optical instruments Manufacture of optical instruments and photographic equipment 332 Manufacture of motor vehicles and their engines 341 Manufacture of bodies (coachwork) for motor vehicles; manufacture of trailers and semi-trailers 342 Manufacture of parts and accessories (auto parts) for motor vehicles and their engines 343 Building and repairing of ships and boats 351 Manufacture of aircraft and spacecraft 353 Manufacture of other transport equipment n.e.c. 359 Manufacture of furniture 361 Manufacturing n.e.c. 369

Source: prepared by the authors, on the basis of the International Standard Industrial Classification of All Economic Activities (isic Rev. 3) adapted for Colombia, as included in the Annual Manufacturing Survey conducted by the National Administrative Department of Statistics (dane) of Colombia. n.e.c.: not elsewhere classified.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 129 0.1 3.2 0.2 2.2 0.0 0.7 4.6 3.4 8.7 7.2 3.9 2.3 0.9 2.1 0.1 1.6 4.1 1.1 12.8 27.1 13.7 Share 2010 ) ( percentages 0.1 3.2 0.2 7.9 1.4 0.0 0.3 4.4 4.7 7.7 7.5 4.6 2.0 1.0 2.3 0.2 1.6 5.1 1.3 28.5 16.0 Share 2000 ) ( percentages 28 457 997 303 644 875 309 188 2010 3 312 162 3 967 757 9 712 592 7 403 344 5 151 153 14 711 050 59 224 740 10 149 219 62 960 026 21 004 502 15 864 573 40 189 407 33 319 757 17 870 744 10 560 992 18 670 111 124 865 969 3 258 18 493 142 964 540 704 512 624 134 435 812 536 2009 2 170 731 8 318 727 1 272 541 8 781 966 2 990 309 1 693 138 4 730 273 4 442 766 2 018 181 1 369 326 1 549 845 1 138 936 2 268 686 19 627 647 5 326 84 705 27 414 109 493 493 701 519 861 983 698 719 666 2008 2 219 561 1 371 686 8 472 098 2 766 491 1 846 630 4 968 192 4 590 775 2 218 742 1 631 329 1 462 750 2 359 955 10 001 524 18 001 494 4 809 73 609 27 181 121 966 540 087 524 803 957 309 545 196 2007 1 850 557 9 413 412 1 887 402 7 215 348 2 780 749 1 951 739 6 888 757 4 535 001 2 088 167 1 296 366 1 245 303 2 205 896 15 224 634 4 421 66 181 13 509 114 961 551 619 468 904 864 007 556 724 2006 8 579 633 1 429 111 6 851 308 2 485 124 1 869 509 5 614 699 3 776 549 2 143 461 1 351 551 1 120 715 2 008 245 1 559 679 13 761 894 3 280 56 619 10 380 103 450 413 447 392 524 932 479 754 396 585 253 2005 6 612 653 1 131 345 6 081 515 2 231 217 1 659 680 4 053 404 2 903 819 1 938 674 1 254 836 1 829 561 1 557 755 11 951 754 dane ) of Colombia. 3 274 53 138 47 234 102 238 913 087 286 143 360 758 832 751 684 379 464 979 2004 5 291 157 5 920 572 1 930 744 1 594 966 3 834 319 3 208 584 1 816 660 1 158 038 1 929 763 1 275 004 10 880 544 0 88 549 41 910 47 995 2003 586 952 161 908 808 361 346 945 747 463 580 462 366 266 3 642 836 9 957 251 5 711 664 1 700 688 1 430 800 3 171 306 3 055 937 1 626 685 1 711 585 1 229 704 1 535 80 937 44 587 47 318 2002 645 634 106 687 578 345 284 834 644 992 517 303 387 859 2 761 632 9 470 550 5 065 855 1 546 592 1 295 873 2 601 694 2 551 045 1 443 904 1 567 476 1 067 624 1 134 74 692 34 478 37 065 2001 539 818 145 158 574 607 285 223 559 232 498 740 367 440 931 382 2 482 134 8 381 305 4 574 087 1 385 012 1 280 019 2 271 306 2 250 868 1 360 344 1 438 933 1 420 58 053 72 707 55 213 32 597 2000 371 644 538 232 271 280 617 062 424 112 345 235 849 054 2 121 031 7 608 895 4 285 615 1 187 576 1 242 219 2 055 458 2 004 414 1 215 924 1 350 012 Industrial sector data Industrial sector data Colombia: value added of industrial sectors, 2000-2010 (Thousands of Colombian pesos and percentages)

Rev. 3) adapted for Colombia, as included in the Annual Annual 3) adapted for Colombia, as included in the ( isic Rev. Activities All Economic prepared by the authors, on basis of International Standard Industrial Classification ANNEX 2 A.2.1 TABLE Source: Department of Statistics ( Administrative conducted by the National Survey Manufacturing Note: share 2000 and share 2010 correspond to the sector’s share of value added in the total industrial value added for 2000 and 2010, respectively. added for 2000 and 2010, respectively. added in the total industrial value share of value Note: share 2000 and 2010 correspond to the sector’s classified. n.e.c.: not elsewhere Manufacture of medical, precision and optical instruments, Manufacture and clocks watches Manufacture of coke, refined petroleum of coke, Manufacture products and nuclear fuel Manufacture of food products and beverages Manufacture Manufacture of motor vehicles, trailers and semi-trailers of motor vehicles, Manufacture Manufacture of chemicals and chemical products Manufacture Manufacture of tobacco products Manufacture Manufacture of other transport equipment Manufacture Manufacture of rubber and plastics products Manufacture Manufacture of textiles Manufacture Manufacture of furniture; manufacturing n.e.c. of furniture; manufacturing Manufacture Manufacture of other non-metallic mineral products Manufacture Manufacture of wearing apparel; dressing Manufacture and dyeing of fur Manufacture of basic metals Manufacture Tanning and dressing of leather; manufacture of footwear; and dressing of leather; manufacture Tanning handbags, accessories, luggage, of travel manufacture saddlery and harness Manufacture of fabricated metal products, except metal products, except of fabricated Manufacture machinery and equipment Processing of wood, manufacture of products of wood and of products wood manufacture Processing of wood, of articles straw furniture; manufacture cork, except and plaiting materials Manufacture of electrical machinery and apparatus n.e.c. Manufacture Manufacture of paper, paperboard and paper of paper, Manufacture and paperboard products Manufacture of motor vehicles, trailers and semi-trailers of motor vehicles, Manufacture Publishing, printing and reproduction of recorded media Manufacture of radio, television and communication of radio, television Manufacture equipment and apparatus

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 130 CEPAL REVIEW 114 • DECEMBER 2014

ANNEX 3 Estimation results

TABLE A.3.1 Colombia: estimation results for equation (3) showing the aggregate impact of the real effective exchange rate on industrial value added Generalized method of moments (gmm) estimates

First-differences System

(1) (2) (1) (2) yt-1 0.603 0.689 0.615 0.658 (0.055)a (0.062)a (0.040)a (0.038)a yt-2 0.023 0.047 0.029 0.049 (0.019) (0.023)b (0.017) (0.020)b wt 0.732 0.758 0.727 0.744 (0.043)a (0.047)a (0.041)a (0.045)a wt-1 -0.402 -0.439 -0.39 -0.395 (0.063)a (0.067)a (0.051)a (0.050)a it -0.102 -0.126 -0.087 -0.118 (0.038)b (0.039)a (0.034)b (0.037)b it-1 -0.295 -0.361 -0.286 -0.360 (0.056)a (0.060)a (0.052)a (0.061)a qt -0.297 -0.269 -0.297 -0.261 (0.087)b (0.0893)a (0.084)a (0.088)b qt-1 0.05 0.165 0.011 0.138 (0.067) (0.081)b (0.064) (0.076)c iitt -1.00E-08 -1.14E-08 -1.12E-08 -1.11E-08 (6.49E-09)a (6.71E-09)c (0.00)a (6.37E-09)c rgdpt 0.022 0.02 (0.024) (0.023) rgdpt-1 -0.529 -0.484 (0.176)b (0.154)b cons 1.806 7.084 6.565 (0.416)a (1.786)a (1.622)a Number of observations 2 906 2 906 3 355 3 355 Observations per group (average) 7.21 7.21 7.712 7.712 Sargan test p-value 0.1404 0.380 0.3047 0.638 c1 (p-value) 0.000 0.000 0.000 0.000 c2 (p-value) 0.2184 0.354 0.2907 0.502

Source: prepared by the authors. a p<0.01. b p<0.05. c p<0.1. Note: standard errors in parentheses. Sargan test for overidentifying restrictions. c1 and c2 tests for first and second order correlation in first-differenced residuals.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 131

TABLE A.3.2 Colombia: estimation results for equation (4) showing the disaggregate impact of the real effective exchange rate on the industrial sectors (1) (2)

Coefficient Standard error Coefficient Standard error a a yt-1 0.579 (0.054) 0.661 (0.061) c yt-2 0.018 (0.018) 0.039 (0.021) a a wt 0.762 (0.041) 0.765 (0.045) a a wt-1 -0.390 (0.059) -0.429 (0.062) rgdpt 0.027 (0.023) c rgdpt-1 -0.344 (0.208) rgdpt-2 0.062 (0.199) b a it -0.109 (0.037) -0.110 (0.038) b a it-1 -0.313 (0.057) -0.331 (0.066) a a iitt -1.160E-08 (6.62E-09) -1.340E-08 (6.870E-09) qt Ds Production, processing and preserving of meat and fish 0.224 (0.482) 0.256 (0.478) Processing of fruits, vegetables, oils and fats -0.013 (0.340) 0.006 (0.324) Manufacture of dairy products -1.142 (0.482)b -0.984 (0.482)b Production of grain mill products, starches and starch products, and prepared animal feeds 0.118 (0.328) 0.240 (0.350) Manufacture of bakery products, macaroni, noodles, couscous and similar farinaceous products -0.085 (0.170) -0.075 (0.199) Manufacture of coffee 0.420 (0.383) 0.618 (0.417) Sugar mills and refineries -1.151 (0.401)a -1.203 (0.412)a Manufacture of other food products -0.279 (0.289) -0.247 (0.262) Manufacture of beverages -0.045 (0.366) 0.154 (0.379) Manufacture of tobacco products 0.299 (0.354) 0.143 (0.473) Preparation and spinning of textile fibres -0.825 (0.521) -0.836 (0.537) Weaving of textiles -0.822 (0.329)b -0.839 (0.343)b Finishing of textiles not produced in the same production unit -1.789 (0.625)a -1.916 (0.703)a Manufacture of other textile products -0.107 (0.494) -0.041 (0.553) Manufacture of knitted and crocheted fabrics and articles 0.424 (0.299) 0.459 (0.313) Manufacture of wearing apparel, except fur apparel 0.506 (0.435) 0.544 (0.455) Tanning and preparation of leather -0.179 (0.288) -0.172 (0.277) Manufacture of footwear -0.538 (0.195)a -0.545 (0.226)b Manufacture of travel goods, handbags and similar articles -0.583 (0.558) -0.603 (0.610) Sawing, planing and impregnation of wood 0.228 (0.802) 0.205 (0.802) Manufacture of veneer sheets, manufacture of plywood, laminboard, particle board and other panels and boards -3.681 (0.214)a -3.810 (0.184)a Manufacture of builders’ carpentry and joinery 0.803 (0.509) 0.835 (0.534) Manufacture of wooden containers -1.258 (0.364)a -1.335 (0.287)a Manufacture of other products of wood, manufacture of articles of cork, straw and plaiting materials -0.357 (0.228) -0.386 (0.180)b Manufacture of paper, cardboard and paper and cardboard products -0.912 (0.385)b -0.850 (0.398)b Publishing 0.679 (0.359)c 0.717 (0.381)c Printing -0.582 (0.257)b -0.675 (0.306)b Service activities related to printing -0.536 (0.955) -0.685 (0.981) Manufacture of refined petroleum products -1.137 (0.542)b -1.125 (0.519)b Manufacture of basic chemicals -0.337 (0.551) -0.146 (0.567) Manufacture of other chemical products -1.033 (0.507)b -0.962 (0.533)c Manufacture of rubber products -1.326 (0.389)a -1.352 (0.332)a Manufacture of plastics products -0.143 (0.168) -0.128 (0.155) Manufacture of glass and glass products 0.428 (0.420) 0.515 (0.441) Manufacture of non-metallic mineral products n.p.c. -1.278 (0.308)a -1.227 (0.315)a Manufacture of basic iron and steel -0.726 (0.335)b -0.791 (0.325)b Manufacture of basic precious and non-ferrous metals -0.636 (0.868) -0.726 (0.955)

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 132 CEPAL REVIEW 114 • DECEMBER 2014

Table A.3.2 (conclusion)

(1) (2)

Coefficient Standard error Coefficient Standard error Manufacture of structural metal products, tanks, reservoirs and steam generators -0.770 (0.378)b -0.819 (0.413)b Manufacture of other fabricated metal products and related metalworking service activities -0.336 (0.382) -0.348 (0.396) Manufacture of general purpose machinery 0.529 (0.425) 0.604 (0.428) Manufacture of special purpose machinery 0.123 (0.285) 0.175 (0.270) Manufacture of domestic appliances n.p.c. 0.048 (0.506) -0.010 (0.547) Manufacture of electric motors, generators and transformers 0.218 (0.891) 0.233 (0.944) Manufacture of electricity distribution and control apparatus -0.588 (0.728) -0.645 (0.748) Manufacture of insulated wire and cable 2.953 (0.148)a 3.013 (0.153)a Manufacture of accumulators and electrical batteries 0.244 (0.561) 0.330 (0.352) Manufacture of electric lamps and lighting equipment -1.033 (0.310)a -1.036 (0.340)a Manufacture of other electrical equipment n.p.c. 0.398 (0.340) 0.405 (0.262) Manufacture of electronic valves and tubes and other electronic components 0.931 (0.615) 0.996 (0.681) Manufacture of television and radio receivers, sound or image recording or reproducing apparatus, and associated goods -4.076 (0.280)a -3.831 (0.323)a Manufacture of medical appliances and instruments and appliances for measuring, checking, testing, navigating and other purposes, except optical instruments -0.981 (0.891) -1.042 (0.972) Manufacture of optical instruments and photographic equipment 0.320 (0.110)a 0.380 (0.094)a Manufacture of motor vehicles and their engines 0.702 (0.646) 0.668 (0.823) Manufacture of bodies (coachwork) for motor vehicles; manufacture of trailers and semi-trailers 0.643 (0.704) 0.622 (0.736) Manufacture of parts and accessories (auto parts) for motor vehicles and their engines -0.349 (0.388) -0.406 (0.450) Building and repairing of ships and boats -3.440 (3.111) -3.409 (3.203) Manufacture of aircraft and spacecraft 0.212 (0.412) -0.272 (0.318) Manufacture of other transport equipment n.p.c. 0.357 (0.261) 0.274 (0.301) Manufacture of furniture 0.338 (0.396) 0.360 (0.408) Manufacturing n.p.c. -0.825 (0.318)a -0.809 (0.330)a _cons 1.812 (0.425)a 4.344 (1.618)a Number of observations 2 906 2 906 Observations per group (average) 7.210 7.210 Sargan (p-value) 0.074 0.144 c1 (p-value) 0.000 0.000 c2 (p-value) 0.096 0.133

Source: prepared by the authors. a p<0.01. b p<0.05. c p<0.1. Note: standard errors in parentheses. Sargan test for overidentifying restrictions. c1 and c2 tests for first and second order correlation in first-differenced residuals. n.p.c.: not previously classified.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. CEPAL REVIEW 114 • DECEMBER 2014 133

TABLE A.3.3 Colombia: estimation results for equation (4) for the sectors significantly affected by the variable (1) (2) qt Ds Coefficient Standard error Coefficient Standard error Manufacture of television and radio receivers, sound or image recording or reproducing apparatus, and associated goods -4.076 (0.280)a -3.831 (0.323)a Manufacture of veneer sheets, manufacture of plywood, laminboard, particle board and other panels and boards -3.681 (0.214)a -3.810 (0.184)a Finishing of textiles not produced in the same production unit -1.789 (0.625)a -1.916 (0.703)a Manufacture of rubber products -1.326 (0.389)a -1.352 (0.332)a Manufacture of non-metallic mineral products n.p.c. -1.278 (0.308)a -1.227 (0.315)a Manufacture of wooden containers -1.258 (0.364)a -1.335 (0.287)a Sugar mills and refineries -1.151 (0.401)a -1.203 (0.412)a Manufacture of dairy products -1.142 (0.482)b -0.984 (0.482)b Manufacture of refined petroleum products -1.137 (0.542)b -1.125 (0.519)b Manufacture of other chemical products -1.033 (0.507)b -0.962 (0.533)c Manufacture of electric lamps and lighting equipment -1.033 (0.310)a -1.036 (0.340)a Manufacture of paper, cardboard and paper and cardboard products -0.912 (0.385)b -0.850 (0.398)b Manufacturing n.p.c. -0.825 (0.318)a -0.809 (0.330)a Weaving of textiles -0.822 (0.329)b -0.839 (0.343)b Manufacture of structural metal products, tanks, reservoirs and steam generators -0.770 (0.378)b -0.819 (0.413)b Manufacture of basic iron and steel -0.726 (0.335)b -0.791 (0.325)b Printing -0.582 (0.257)b -0.675 (0.306)b Manufacture of footwear -0.538 (0.195)a -0.545 (0.226)b Manufacture of optical instruments and photographic equipment 0.320 (0.110)a 0.380 (0.094)a Publishing 0.679 (0.359)c 0.717 (0.381)c Manufacture of insulated wire and cable 2.953 (0.148)a 3.013 (0.153)a

Source: prepared by the authors. a p<0.01. b p<0.05. c p<0.1. Note: we present in this table the significantly estimated results for the parameterb 2 in equation (3). Standard errors in parentheses. The shaded rows correspond to the sectors that have been affected positively by the real appreciation of the Colombian peso. n.p.c.: not previously classified.

TABLE A.3.4 Colombia: share of value added, employees and firms of the sectors that saw a significant impact on value added as a result of real exchange rate appreciation

Sectors significantly impacted by real Percentages of total exchange rate appreciation Value added Number of employees Number of firms Negative effect 52.7 38.7 35.9 Positive effect 4.0 2.1 5.7

Source: prepared by the authors.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. 134 CEPAL REVIEW 114 • DECEMBER 2014

Bibliography Acosta, P., E. Lartey and F. Mandelman (2009), “Remittances and Corden, W.M. and J.P. Neary (1982), “Booming sector and de- the Dutch disease”, Journal of International Economics, industrialisation in a small open economy”, Economic Journal, vol. 79, No. 1, Amsterdam, Elsevier. vol. 92, No. 368, Hoboken, John Wiley & Sons. Algieri, B. (2011), “The Dutch disease: evidences from Russia”, Dinero (2011), 18 August [online] http://www.dinero.com/pais/articulo/ Economic Change and Restructuring, vol. 44, No. 3, Springer. santos-reune-empresarios-inicio-su-visita-argentina/133589. Anderson, T. and C. Hsiao (1981), “Estimation of dynamic Echavarría, J.J. and M.A. Arbeláez (2003), “Tasa de cambio y models with error components”, Journal of the American crecimiento económico en Colombia durante la última Statistical Association, vol. 76, No. 375, Alexandria, American década”, Borradores de Economía, No. 338, Bogota, Bank Statistical Association. of the Republic. Arellano, C. and others (2005), “The dynamic implications of foreign Echavarría, J.J. and M. Villamizar (2006), “El proceso colombiano aid and its variability”, imf Working Papers, No. 05/119, de desindustrialización”, Borradores de Economía, No. 361, Washington, D.C., International Monetary Fund, June. Bogota, Bank of the Republic. Arellano, M. and O. Bover (1995), “Another look at the instrumental Egert, B. and C. Leonard (2008), “Dutch disease scare in Kazakhstan: variable estimation of error-component models”, Journal of is it real?”, Open Economies Review, vol. 19, No. 2, Springer. Econometrics, vol. 68, No. 1, Amsterdam, Elsevier. Ekholm, K., A. Moxnes and K. Ulltveit-Moe (2012), “Manufacturing Arellano, M. and S. Bond (1991), “Some tests of specification restructuring and the role of real exchange rate shocks”, for panel data: Monte Carlo evidence and an application to Journal of International Economics, vol. 86, No. 1, employment equations”, Review of Economic Studies, vol. 58, Amsterdam, Elsevier. No. 2, Wiley Blackwell. Goldberg, L.S. (1993), “Exchange rates and investment in United Berg, A., J. Ostry and J. Zettelmeyer (2012), “What makes growth States industry”, The Review of Economics and Statistics, vol. sustained?”, Journal of Development Economics, vol. 98, 75, No. 4, Cambridge, Massachusetts, The mit Press. No. 2, Amsterdam, Elsevier. Goldberg, L., J. Tracy and S.R. Aaronson (1999), “Exchange rates Berman, N., P. Martin and T. Mayer (2012), “How do different and employment instability: evidence from matched cps exporters react to exchange rate changes?”, The Quarterly data”, American Economic Review, vol. 2, No. 89, Nasvhille, Journal of Economics, vol. 127, No. 1, Oxford University Press. Tennessee, American Economic Association. Beverellli, C., S. Dell’Erba and N. Rocha (2011), “Dutch disease Guha, P. (2013), “Macroeconomic effects of international remittances: revisited. Oil discoveries and movements of the real exchange the case of developing economies”, Economic Modelling, rate when manufacturing is resource-intensive”, International vol. 33, Amsterdam, Elsevier. Economics and Economic Policy, vol. 8, No. 2, Springer. Lartey, E.K.K. (2011), “Financial openness and the Dutch Buiter, W.H. and D.D. Purvis (1983), “Oil, disinflation, and export disease”, Review of Development Economics, vol. 15, No. 3, competitiveness: A model of the Dutch Disease”, Economic Wiley Blackwell. Interdependence and Flexible Exchange Rates, J.S. Bhandari Magud, N. and S. Sosa (2010), “When and why worry about real and B.H. Putnam (eds.), Cambridge, Massachusetts, The exchange rate appreciation? The misssing link between mit Press. Dutch disease and growth”, imf Working Papers, No. 10/271, Burgess, S. and M. Knetter (1998), “An international comparison of Washington, D.C., International Monetary Fund, December. employment adjustment to exchange rate fluctuations”, Review Poncela, P., E. Senra and L. Sierra (2012), “Is the boost in oil prices of International Economics, vol. 6, No. 1, Wiley Blackwell. affecting the appreciation of real exchange rate?: empirical Campa, J. and L. Goldberg (2001), “Employment versus wage evidence of ‘Dutch disease’ in Colombia”, Documentos de Trabajo, adjustment and the us dollar”, Review of Economics and No. 694, Madrid, Fundación de las Cajas de Ahorros (funcas). Statistics, vol. 83, No. 3, The mit Press. Prati, A. and T. Tressel (2005), “Aid volatility and Dutch disease: (1995), “Investment in manufacturing, exchange rates and is there a role for macroeconomic policies?”, imf Working external exposure”, Journal of International Economics, vol. 38, Papers, No. 06/145, Washington, D.C., International Monetary No. 3-4, Amsterdam, Elsevier. Fund, March. Campbell, J.R. and B. Lapham (2004), “Real exchange rate Rajan, R.G. and A. Subramanian (2011), “Aid, Dutch disease, and fluctuations and the dynamics of retail trade industries on the manufacturing growth”, Journal of Development Economics, U.S.-Canada border”, The American Economic Review, vol. 94, vol. 94, No. 1, Amsterdam, Elsevier. No. 4, Nashville, Tennessee, American Economic Association. Rhenals, R. and J. Saldarriaga (2007), “Tasa de cambio real y Carranza, J. and S. Moreno (2013), “Tamaño y estructura vertical crecimiento económico en Colombia, 1995-2006: Una de la cadena de producción industrial colombiana desde exploración inicial”, Perfil de Coyuntura Económica, Medellin, 1990”, Borradores de Economía, No. 751, Bogota, Bank of Antioquia, University of Antioquia. the Republic. Rodrik, D. (2008), “The real exchange rate and economic growth”, Clavijo, S. (1990), “Productividad laboral, multifactorial y la tasa de Brookings Papers on Economic Activity, vol. 39, No. 2, cambio real en Colombia”, Ensayos sobre política económica, Washington, D.C., The Brookings Institution. Bogota, Bank of the Republic. Salter, W. (1959), “Internal and external balance: the role of price and Corden, W.M. (1984), “Booming sector and Dutch disease economics: expenditure effects”, Economic Record, vol. 35, No. 71, Wiley. Survey and consolidation”, Oxford Economic Papers, vol. 36, Swan, T. (1960), “Economic control in a dependent economy”, No. 3, Oxford, Oxford University Press. Economic Record, vol. 36, No. 73, Wiley.

A FIRST APPROACH TO THE IMPACT OF THE REAL EXCHANGE RATE ON INDUSTRIAL SECTORS IN COLOMBIA • LYA PAOLA SIERRA AND KARINA MANRIQUE L. Global integration, disarticulation and competitiveness in Mexico’s electromechanical sector: A structural analysis

Raúl Vázquez López

ABSTRACT This article analyses the dual functioning of the Mexican electromechanical sector between 1994 and 2008, as distinct from other globalized activities. An estimation of labour productivity in 52 industrial classes finds that structural heterogeneity increased particularly in the 1994-2001 subperiod, alongside technical and organizational improvements that were increasingly concentrated in a small number of subsidiary companies of transnational automotive-assembly enterprises. The application of a shift-share technique also revealed the absence of any significant structural change. Lastly, an extension of the methodology to evaluate competitiveness —developed by the Economic Commission for Latin America

and the Caribbean (eclac)— and its application to a second database that reclassifies 1,345 foreign trade products, makes it possible to contrast these changes with the dynamism of the global production networks in which the leading firms of the sector in Mexico are engaged.

KEYWORDS Engineering industries, industrial organization, production specialization, productivity, competitiveness, evaluation, Mexico

JEL CLASSIFICATION F68, L16, L62

AUTHOR Raúl Vázquez López is a Level “B” Tenured Research Fellow at the Institute of Economic Research of the National Autonomous University of Mexico (unam), Industrial Economics Unit. [email protected] 136 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

Since the 1980s, Mexico and several other Latin American to the rest of the economy, and thus help to “lift up” countries have been implementing an economic- backward populations, areas or sectors (Pinto, 1970, development strategy based on open trade and deregulation p. 97)). This same historical experience shows that of the economy. The objectives of the measures thus structural change based on the development of adopted —inspired in an orthodox vision that believes complementarities between the different activities the market has autonomous capacity to allocate resources requires participation by a complex institutional fabric.1 efficiently— include promoting structural change in the Pursuing these ideas, the present article seeks productive system by integrating the model’s leading to analyse the structural evolution of the Mexican activities into global value chains. Backdropped by a electromechanical sector, leader of the country’s sophisticated international division of labour, the strategy manufacturing export profile, highlighting the dualism that based on productive specialization assumes that freely exists between globalized and non-globalized activities. functioning markets, supported by the adopted measures, For this purpose, two databases were constructed at enable jobs to be created and factors to be reallocated the highest possible level of disaggregation for the towards the most productive uses. period 1994-2008. The first refers to levels of labour Three decades on, the results obtained have called productivity in 52 classes of activity in the sector; and the validity of this argument into question. In Mexico, the second relating to the exports and imports of the public policy has adhered to orthodox guidelines in different sector activities (see annex) —resulting from exemplary fashion. Trade liberalization and economic the reclassification, nonexistent at the time, of 1,345 deregulation have continued apace, while maquila activity foreign trade products defined by the 1992 Harmonized has experienced an unprecedented boom. Capitalizing System (HS92) in terms of the Mexican Activities and on the advantages afforded by geographical proximity Products Classification (cmap). The core of this study to the United States market and the signing of the presents the results of different statistical exercises based North American Free Trade Agreement (nafta), large on these information sources. transnational corporations have invested in the country; The first part of section II, which is theoretical, and the growth of exports of Mexican-assembled products discusses the role played by public policy in integrating has outpaced world trade as a whole. Nonetheless, the the leading Mexican manufacturing industries into global external competitiveness of the model’s leading industries production networks; and the second part documents has been unable to counteract the deindustrialization how the different activities’ shares in the output and process in non-globalized subsectors of the economy, employment of the electromechanical sector have which have to compete in the domestic market with become more concentrated. Section III considers the imports that benefit from the adopted measures. heterogeneous pattern of change in disaggregated levels From the standpoint of the theoretical tradition of of labour productivity, while section IV uses a generic Latin American structuralist thinking, the spillover effects shift-share technique to distinguish the main determinants from the most efficient industries in terms of transferring of the changes observed. Lastly, after section V has technological capacities, modes of organization, and evaluated the sector’s external competitive performance demand for inputs, are fundamental for achieving a in dynamic terms, along with its constraints, section VI major transformation of the productive sector. This sets forth the main conclusions. would underpin not only on economic growth but also development in a broader sense (when considering the historical experience of the developed countries in 1970, 1 José Antonio Ocampo, former Executive Secretary of eclac, defines Aníbal Pinto proposed as a key development objective for the concept of complementarities extensively, referring not only to the role of the backward and forward linkages identified by Hirschman, modern industries, which he defined as those of higher but also to the role played by public, private, or mixed institutions productivity, that they should transmit their progress created to reduce information costs (Ocampo, 2005).

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 137

II Public policy in Mexico and global production networks

Arndt and Kierzkowski (2001) relate the international those local capacities to complement the strategies fragmentation of production to the emergence of assembly defined by transnational actors located within these or maquila activities in less developed countries; and global production networks. In short, the transnationals they characterize this phenomenon in terms of previously condense a systemic power which they exercise under integrated productive processes being split into one or profit-maximization criteria, so as to transform the more components, the manufacturing of which migrates national and subnational regulatory frameworks and, geographically, thereby giving rise to intra-industry and definitively, the productive structures they subordinate intra-product trade. Yamashita (2008) makes clear that (Dawley, 2011). Some discussions of this even define this cross-border division of the productive process “corporate capture” as the potential for transnational leads to activity segments that make intensive use of enterprises to harness institutional capacities to the low-skill labour being relocated towards developing detriment of the interests of national firms and workers countries, while tasks with a high knowledge content (Phelps, 2008). or those involving sophisticated technologies are kept in advanced countries. 1. Integration of the leading Mexican The above results in a highly hierarchical and rigid manufacturing industries into global global form of organization, in which the benefits are production networks unequally distributed, entailing a twofold specialization of the national economies. As capital-intensive productive The foregoing is relevant mainly in terms of the role segments would thus fall outside the “specialization of public institutions in developing countries, such as cone” of labour-abundant developing countries, tasks Mexico, which are specialized in tasks involving the that involve higher technology and knowledge content assembly of manufactured goods. The promotion of not only do not migrate towards these countries, but duty-free importation of components, intermediate actually disappear from them if they previously existed goods and inputs, along with the acceptance by the (Deardoff, 1979). By testing the theory against empirical national authorities of a value-added tax imposed on the evidence for the Mexican manufacturing sector, Puyana product to be re-exported to the country of origin, have and Romero argue that this phenomenon would explain aimed to underpin the organizational strategies of these the reduction in the national content of certain activities global production networks to the detriment of local —in the automotive sector for example— which productive linkages and income levels (Yeats, 2001). occurred after the Mexican economy was liberalized In reality, the public-policy measures implemented by in the wake of the debt crisis (Puyana and Romero, developing countries, to attract foreign investment and 2006, p. 72). maquila activities, have mostly served as a disincentive By introducing elements of geographical political to national content in the manufacturing process; and economy into the analysis of global production networks, they are ultimately the result of needs created by the MacKinnon (2012) finds that the role of institutions competitive pressures experienced by the parent companies has been to ensure strategic coupling2 between locally of large transnational enterprises in advanced countries existing potentials and the needs posed by the firms that (Arndt, 2001). drive these international networks. Coe and others (2004) In the case of Mexico, Puyana and Romero (2006) also mention the role of these institutions in moulding document a maquila “bonanza” stemming from the tax incentives provided by the governments of Mexico and the United States, which aimed at reducing costs 2 Yeung (2009, p. 213) defines strategic coupling as the dynamic of production, enhancing profitability, and simulating process through which economic agents in either cities or regions, or both, coordinate, mediate, and arbitrate strategic interests between investments in maquila activity, with a consequent shift local stakeholders and their counterparts in the global economy. of productive factors towards that activity. Under this

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 138 CEPAL REVIEW 114 • DECEMBER 2014 arrangement, the small proportion of national value-added the domestic autoparts industry. The suppression of the incorporated in the exported goods is explained by the Automotive Decree eliminated the following regulatory combined effect of the incentives for duty-free importation provisions: the upper limit on foreign capital in enterprises of components, the charging of a tax on Mexican value- making vehicle parts (originally 49%); the domestic added in the United States, the global rationale underlying value-added requirement in the output of assembly the fragmentation of productive processes that restricts plants (set at 60% prior to 1998); and the minimum the development of capital and knowledge-intensive export value required for every dollar of imports (prior activities, and exchange-rate revaluation which raises to nafta, US$ 1.75 of exports was required for every the relative cost of domestic inputs. dollar of imports) (Hernández, 2000). Deregulation In this sense, public policy has been heavily biased thus meant a significant reduction in the number of in favour of the interests of transnational players, pursuant tasks undertaken, as determined by the productive to a neoliberal economic strategy founded on financial needs of the integrated global system and consisting deregulation and trade liberalization. In 1996, the Industrial of greater specialization in the final assembly-related Policy and Economic Deregulation Programme was manufacturing segments. absorbed by the Foreign Trade and Export Promotion Programme, in the belief that promotion measures in a 2. Concentration and disarticulation in the context of globalization should not be separated from Mexican electromechanical sector those related to international trade (Hernández, 2000). The resulting programme, which provided a frame of In the case of the Mexican electromechanical sector, reference for national industrial policy, then favoured the strategic coupling between the manufacturing structure export sector ahead of the manufacture of nontradable and the needs of global production networks, as defined goods, with the specific central aim of enhancing the by the parent companies of transnational enterprises, competitiveness of the productive structure and helping has mainly manifested itself since trade liberalization it to integrate into global production networks. in a growing divergence between the characteristics A trustworthy indication of the authorities’ of industries serving the domestic market and those responsibility in the transformation of manufacturing that are integrated into international chains generating industry was the implementation of specific promotion exports. In order to obtain an initial differentiated programmes centred on two main lines of action: export approach to the sector’s evolution, a database was promotion (ecex and altex)3 and the development constructed primarily to estimate labour-productivity of maquila activity (pitex, inmex and drawback).4 levels, separating the surveyed classes into two groups Key provisions of these latter programmes include tax (see annex). An initial group covers activities associated exemptions on the temporary importation of intermediate with the automotive industries and those relating to the goods and inputs used in the manufacture, processing electrical-electronic subsector and the manufacture of or repair of export merchandise (general import duty, domestic and office equipment, hereinafter identified value added tax and, where appropriate, countervailing as the “globalized group”; whereas the second group duties). The industrial specialization pattern defined by encompasses other activities and is referred to as the adopted measures thus favoured a subordinate role for “non-globalized.” the national productive apparatus in globalized sectors In general, the data for the period under analysis controlled by the “governance” of global production (1994-2008) show that the total sector shares of both networks (Vázquez, 2012). groups stagnated, both in terms of output value and In the automotive sector, which is a pillar of the in terms of man-hours worked. This would suggest growth strategy, the State progressively dismantled that open trading arrangements failed to generate any legislation which had hitherto proven successful in significant structural change that increased the weight terms of its capacity to encourage exports and develop of activities involved in international chains in the domestic manufacturing structure. Nonetheless, a more disaggregated analysis calls this statement into question, 3 Foreign Trade Enterprises Programme (ecex) and Highly Exporting since a single activity “Manufacture and assembly of Enterprises Programme (altex). cars and trucks” (class 384110), which has the highest 4 Temporary Import Programme for Export (pitex); Programme of level of global integration, increased its share of output the Manufacturing, Maquila and Export Service Industry (immex), and the Programme of Import Duty Refunds to Exporters (drawback). value from 40.8% in 1994 to 50.4% in 2008. In this

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 139 latter year it accounted for half of the sector’s total TABLE 2 output, significantly more than offsetting the fall in Coefficients of articulation and integration of the industry groups in 2003a the share of its main input supplier, “Manufacture of (Percentages) motors and their parts for automobiles and trucks” (class 384122), from 10.6% in 1994 to 6.6% in 2008 Coefficient of Coefficient of Groups (see table 1). articulationb integrationc Globalized group 44.6 56.7 Non-globalized group 61.5 67.8 TABLE 1 Total sector 46.6 58.5 Share of the industry groups in the total gross production of the sector and Source: prepared by the author, on the basis of data from the Herfindahl-Hirschman indices hhi( n), National Institute of Geography and Statistics (inegi), “Encuesta 1994 and 2008 Industrial Anual. Clasificación Mexicana de Actividades y Productos (Percentages) (cmap), 205 clases de actividad económica”, 2013. a The data were calculated for 2003 because this was the last Groups 1994 2008 year for which the necessary information exists under the cmap classification. Globalized group 80.1 81.8 b Coefficient of articulation: national raw materials and auxiliary Activity class 384110 40.8 50.4 inputs consumed/total raw materials and auxiliary inputs consumed. Activity class 384122 10.6 6.6 c Coefficient of integration: (value-added + national raw materials Non-globalized group 19.9 18.3 and auxiliary inputs consumed)/total gross production. Sector total 100 100 hhin globalized group 41.3 52.9 hhin non-globalized group 7.5 12.1 hhin total sector 34.1 43.8 In this regard, the process of disarticulating local value chains in the sector goes hand-in-hand with a Source: prepared by the author, on the basis of data from the National Institute of Geography and Statistics (inegi), “Encuesta progressive concentration of production, mainly in a single Industrial Mensual (eim). Clasificación Mexicana de Actividades activity controlled by the subsidiaries of large foreign y Productos (cmap), 205 clases de actividad económica”, 2013. transnational enterprises (the labour-intensive assembly of automobiles and trucks).6 This is corroborated by calculating the normalized Herfindahl-Hirschman index The diametrically opposing trends experienced (hhin),7 since the indicator rises in all cases, and by a by the two main classes of the sector, integrated into considerable amount particularly in the globalized group, the same subsector and value chain, reflect the fact that from 41.3% in 1994 to 52.9% in 2008 (see table 1). Thus imports have displaced locally supplied vehicle parts, thereby breaking domestic productive linkages headed by the industry leaders engaged in global production total gross production, with respect to each of the 53 activity classes networks. On this point, in the context of the globalization in the sector, and for the groups considered. The information source of the automotive sector, Álvarez states that “local firms used in this case was the Annual Industrial Survey, 205 activity classes have ceased to be suppliers of the assembly plants, (cmap), conducted by inegi (2013b). 6 deferring to the new foreign firms or else engaging in The total output of Mexico’s automotive sector in 2008 amounted to 2,180,294 units, of which 76.4% were destined for the international the importation and distribution of autoparts” (Álvarez, market. In the case of automobiles, production in that year totalled 2002, p. 46). This hypothesis is confirmed by comparing 1,387,913 units, of which 79.5% were exported; 32.4% were produced the results obtained from a calculation of the coefficients by the German Volkswagen assembly unit, 28.3% by the Japanese firm Nissan, and 19.7% and 14.8% by the United States Enterprises of articulation and integration of the globalized and Ford and General Motors, respectively (inegi, 2010). non-globalized groups: whereas the first indicator is 7 The normalized Herfindahl-Hirschman index was separately calculated 16.9% lower for the globalized group, the second is for production values and man-hours worked in the 23 activity classes 5 of the globalized group and in the 30 classes in the non-globalized 11% lower (see table 2). group, and also for all 53 classes of the sector. The formula used was:

n 1 P2 − < / i = 1 i n F 5 HHIn = # 100 Following a review of the information available in the various statistical 1 sources, the coefficient of articulation is defined in this study as the 1 − n value of domestic raw materials and auxiliary inputs consumed, as a percentage of the total value of those inputs consumed. The coefficient where Pi = Xi / Xt indicates the share of class i in the total production of integration was calculated as the sum of value-added and the value value or man-hours worked of the group in question (formula normalized of national raw materials and auxiliary inputs consumed in relation to on the basis of Durán and Álvarez, 2008).

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 140 CEPAL REVIEW 114 • DECEMBER 2014 there is evidence of a strong correlation between a trade uses. This feature would therefore imply that it would liberalization process, driven by a public policy aimed be impossible for such firms to drive any significant at integrating the national productive system into global structural change —a hypothesis that will be evaluated production networks, and a productive specialization in the following sections by analysing trends of labour trend that concentrates activity in a small number of productivity and its determinants. industries and firms within those networks. It should also be noted that, owing to the high TABLE 3 technological level of manufacturing processes in Share of the industry groups in man-hours the globalized activity classes (as exemplified by the worked in the sector and normalized automotive sector), both the increase in these classes’ Herfindahl-Hirschman indices hhi( n), total sector shares, and their degree of concentration, 1994 and 2008 (Percentages) are smaller in terms of man-hours worked. Comparing the situation in 2008 with that prevailing in 1994, the Groups 1994 2008 share of the globalized group in total time worked in the Globalized group 60.2 62.4 sector increases by just 2.2%, whereas the group’s index Activity class 384110 12.5 13.2 of concentration shows a residual growth of 1.6% (see Activity class 384122 8.3 8.0 Non-globalized group 39.8 37.6 table 3). Consequently, an element that would explain Sector total 100 100 the lack of spillover from exports to economic growth hhi globalized group 12.2 13.8 and employment is the inability of these globalized firms hhi non-globalized group 5.6 6.7 hhi total sector 7.8 9.3 to generate jobs on a scale that reflects their standing in the productive structure. This contradicts the orthodox Source: prepared by the author, on the basis of data from the National Institute of Geography and Statistics (inegi), “Encuesta theoretical claim that liberalization and the market Industrial Mensual (eim). Clasificación Mexicana de Actividades alone can reallocate labour towards the most profitable y Productos (cmap), 205 clases de actividad económica”, 2013.

III Trend of labour productivity and structural heterogeneity

In general, the estimations made from the constructed and jointly accounted for 23.2% of the sector’s total database show labour productivity in the period (1994- exports,8 their labour productivity declined sharply 2008) growing at very similar rates in the two industry at rates of 32.5% and 47.3% between 1994 and 2008 groups (globalized and non-globalized), and in line with (see table 4). Consequently, the aggregate performance the trend of that indicator both with respect to the sector of the globalized group is not visibly superior to that total and for manufacturing as a whole (see table 4). of the non-globalized group, and there are signs of In the case of the globalized group, seven of the 23 a “spurious” competitiveness that is not based on classes report a reduction in their labour-productivity technological and organizational improvements in certain level from 1994 to 2008, particularly in the classes highly-exporting activities. “Manufacturing, assembly and repair of communication, The sector’s key activity, class 384110 (Manufacture transmission and signaling equipment” (383201) and and assembly of cars and trucks), records both the highest “Manufacture and assembly of radios, television receivers level of labour productivity of the sample in 2008, and and audio equipment” (383204), which have substantial a well-above- average rise in that indicator between export activity in a national electronics industry that relies heavily on the functioning of global production networks. Despite the competitive success of these 8 Exports in class 383201 correspond to a combination of the following HS92 categories: 8517+8521+8525+8526+8530+8531- activities, which generated external sales of US$ 14,407 851790-853090-853190. In the case of class 383204, the identity is million and US$ 24,999 million, respectively, in 2008 8518+8519+8520+8527+8528-851850 (see annex).

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 141

1994 and 2008. In contrast, the labour productivity real terms; and, although time worked decreased in 10 of the sector’s main input supplier, “Manufacture of of the group’s activities, 10 classes recorded declining car and truck motors and their parts” (class 384122), labour productivity during the period studied). represents just 21.4% of the productivity level of the Meanwhile the standard deviation of the indicator terminal industry having weakened further in the for the globalized group and for the whole sector rises period studied (see table 4). On this point, the analysis by 52.2% when comparing 2008 with 1994, whereas the of all the estimations performed points to widening coefficient of variation rises by 23.2% for the globalized efficiency gaps both between the leading and backward group and by 20.6% for the sector. Nonetheless, there activities of the sector, and also within industries that are are two clearly differentiated subperiods. In the first one integrated into global production networks, and especially (1994-2002), following the entry into force of nafta between those grouped in the input-supplier classes and and a rapid process of trade and financial liberalization assembly industries located in the final phases of the in the Mexican economy, structural heterogeneity in value chain. the sector increases quickly, but to a greater extent within the globalized group, which includes the export industries that are privileged by the new public-policy TABLE 4 Trend of labour productivity by industry measures (annual average growth rates of the standard groups, 1994 and 2008 deviation and coefficient of variation are 8.3% and (Mexican pesos at December 2003 6.4%, respectively, for the sector as a whole; 9.7% and prices and percentages) 6.7% for the globalized group, and 3.7% and 3.3% Groups 1994 2008 Growth rate for the non-globalized group). In a second period of Globalized group 691.8 990.2 43.1 relative stabilization, and following the shakeout of Activity class 384110 1 697.7 2 892.4 70.4 much of the national productive apparatus, the trend, Activity class 384122 664.0 619.4 -6.7 once again led by the globalized firms, reverses after Activity class 383201 744.7 502.5 -32.5 Activity class 383204 699.3 368.5 -47.3 peaking in 2002 to post negative annual average of Non-globalized group 259.1 366.1 41.3 rates growth of the standard deviation and coefficient Sector total 519.4 755.3 45.4 of variation are negative between 2002 and 2008 (-4.1% Manufacturing total 446.1 625.0 40.1 and -4.4 respectively for the sector as a whole, and Source: prepared by the author, on the basis of data from the -2.6% and -4.5 for the globalized group (see figures National Institute of Geography and Statistics (inegi), “Encuesta 1 and 2)). Industrial Mensual (eim). Clasificación Mexicana de Actividades y Productos (cmap), 205 clases de actividad económica”, 2013. Various factors help explain the widening gaps in labour productivity between the activity classes of the electromechanical sector and their timings in According to the theoretical approach adopted Mexico. Firstly, following trade liberalization, the in this study, a structural homogenization process is global integration process in the leading industries a pre-requisite for advancing towards a more mature of the sector specialization pattern generated higher forms of industrialization (Furtado, 1967; Pinto, 1965 degrees of structural heterogeneity, particularly within and 1970), so the widening of productive gaps within the globalized group. The entry or takeoff of a few the leading industries of the export specialization profile large and technically advanced exporting firms, mostly entails an involution of the manufacturing structure. An subsidiaries of foreign multinationals, coexisted with the initial approach to the topic of structural heterogeneity incapacity of many other local industries to be competitive in the Mexican metalwork and machinery sector seems internationally, and with the aforementioned displacement to confirm this hypothesis. The traditional dispersion of local supplies by imports. This was stimulated and statistics for the indicator, calculated both by activity protected by national export-promotion programmes that class and with respect to the groups defined above facilitated the purchase abroad of the parts and inputs (globalized and non-globalized), rise all cases. The to be incorporated into the exported goods. On this largest increase occurs in the non-globalized group as point, Unger stated that “Mexican industry consolidates result of a wide variety of patterns of investment and increasingly concentrated oligopolistic structures, owing technological change, and where there is a predominant to the influence of the large transnational enterprises, phenomenon of obsolescence and deindustrialization and large export-oriented domestic conglomerates” (in 16 of the 30 classes considered, output declines in (Unger, 2001, p. 100).

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 142 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 1 Trend of the standard deviation of labour productivity by activity classes and industry groups, 1994-2008 (Mexican pesos at December 2003 prices)

720

670 Globalized group 620

570

520 Sector total 470

420 Pesos 370

320

270 Non globalized group 220

170

120 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Source: prepared by the author, on the basis of data from the National Institute of Geography and Statistics (inegi), “Encuesta Industrial Mensual (eim). Clasificación Mexicana de Actividades y Productos (cmap), 205 clases de actividad económica”, 2013.

FIGURE 2 Trend of the coefficient of variation of labour productivity by activity classes and industry groups, 1994-2008 (Coefficients calculated on the basis of figures expressed in Mexican pesos at December 2003 prices)

1.30

1.20

1.10 Sector total 1.00 Globalized group 0.90

0.80

0.70 Non-globalized group

0.60

0.50 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Source: prepared by the author, on the basis of data from the National Institute of Geography and Statistics (inegi), “Encuesta Industrial Mensual (eim). Clasificación Mexicana de Actividades y Productos (cmap), 205 clases de actividad económica”, 2013.

Similarly, some activity branches were restructured sized enterprises (smes) and entire industrial segments to meet the requirements defined by the “governance” of collapsed and disappeared, leading to a shakeout of global production networks, and to be able to undertake the national productive apparatus. In a synthesis of this tasks with different factor contents and essentially transformation, Capdevielle notes that “the composition assembly work. As a result of competition from imports of manufacturing output changed, partly because the on the local market, most of the small and medium- economic sectors that produced for export were specialized

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 143 in certain segments of the value chain, and, at the same as there was no flow of knowledge outside the global time, local production was displaced by imports, thanks network, nor any significant positive externalities, and no to liberalization and the availability of foreign exchange generalized organizational learning, the efficiency gaps obtained from the new form of participation in trade” continued to widen. This happened even more quickly (Capdevielle, 2005, p. 108). in the residual industries of the non-globalized group, It is in this new context of high market concentration, which were undergoing a genuine de-industrialization and in a second period (2003-2008), that degrees of process. It should be noted that this periodization is heterogeneity within the globalized group decrease, with consistent with the findings of previous studies for the presence of limited dissemination of technology and Mexican manufacturing as a whole, albeit with contrasting capacities within the leading activities. At the same time, trends in the subperiods considered (Vázquez, 2013).

IV Deindustrialization and structural change

In the context of the deindustrialization of activities between 1994 and 2008. Needless to say, the effects may serving the domestic market, two elements tend to be negative when factors of production shift towards corroborate the hypothesis put forward above in lower-productivity activities (structural effect), or when relation to the absence of a reinvigorating change in labour productivity declines owing to technological the overall structure of the Mexican metalwork and obsolescence or outdated modes of organization within machinery sector: firstly, greater heterogeneity in terms the different activities (intrinsic effect). of technologies and capacities; and, secondly, increased The mathematical formulation of this decomposition, concentration of productive capacity, both in a small which is applied to compare the indicator values for number of classes (especially those related to assembly the 53 classes at two points in time (1994 and 2008) activities) and, within those classes, in a few large firms is as follows: that are integrated into global production networks and 2008 1994 benefit from the public-policy measures implemented PP− n _ i in the wake of trade liberalization. Consequently, this =−PP2008 1994. SS 1994+ 2008 2 (1) /i = 1 ii ii section seeks to evaluate the determinants of the labour- :``jjD n productivity trends observed in the different activities, +−SS2008 1994. PP 1994+ 2008 2 /i = 1 : ii iiD to confirm the concentration of efficiency gains in a few ``jj globalized classes and the very weak role played in the where Pi is productivity in activity i (i = 1, 2, ...n) improvements by the migration of labour towards more and Si is the share of the activity i (i = 1, 2, ...n) in the efficient uses. total active population employed in the sector. The first A standard methodology of the shift-share type, term on the right-hand side of equation (1) represents the a descriptive technique commonly found in this type variation in labour productivity caused by changes in the of study, was used to identify the components of intrinsic productivity of the n activity classes (intrinsic productivity changes (total effect) at two points in effect). The second term identifies the contribution made time. The first component was an effect related to by the recomposition of the labour force (structural the structural change, in other words the migration of effect) (eclac, 2007). productive factors towards more efficient uses (structural The results of the exercise are consistent with the effect); and the second identified an effect linked to the analysis performed previously. In the subperiod 1994- changes that occurred within each activity, which can 2008, the growth of labour productivity in automotive be associated with technical progress (intrinsic effect). assembly (class 384110) represents 83.7% of that Using the information base described in the first section recorded in the globalized group altogether, and 71.4% of this article, the following paragraphs break down the of the increase in the indicator for the entire sector. increases in labour productivity that occurred in the sector Of the contribution made by this activity, the intrinsic

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 144 CEPAL REVIEW 114 • DECEMBER 2014 effect, attributable to technological and organizational within the activity segment, leading to a widening of improvements within the class, explains 90.9% of that efficiency gaps between the terminal industry and its improvement, while the structural effect on the progress potential local suppliers (see table 5). achieved is residual. Meanwhile, in activities related to The non-globalized group makes only a small the manufacture of inputs for automotive assembly, the contribution to labour-productivity growth in the sector modernization process is weak to say the least (the sum during the period, accounting for just 14.6% of the of the intrinsic effects amounts to 23 Mexican pesos total sector effect valued at 235.9 Mexican pesos at at December 2003 prices in the period considered).9 December 2003 prices. Within this group, however, Moreover, in class 384122 (supply of motors and parts), efficiency improvements are highly concentrated. The the two effects as calculated are actually negative, thereby sum of classes 381412: “Electroplating of metal parts”, denoting a lack of technology diffusion and capacities and 382206: “Manufacture of air-conditioning machines and refrigeration and heating equipment”, accounts for 59.8% of the rise in the indicator in the non-globalized 9 Results from the sum of the intrinsic effects relating to classes group. Within this group, the results of the activity 384121, 384122, 384123, 384124, 384125 and 384126. “Manufacture of containers and products made of tin,

TABLE 5 Determinants of the trend of labour productivity by activity classes and industry groups 1994-2008 (Mexican pesos at December 2003 prices)

Non-globalized group Globalized group

Intrinsic Structural Intrinsic Structural Activity class Total effect Activity class Total effect effect effect effect effect

381100 5.1 -0.1 5.0 384110 153.1 15.3 168.5 381201 3.9 -3.7 0.2 384121 12.2 -2.5 9.8 381202 -0.4 -0.8 -1.2 384122 -3.6 -1.8 -5.4 381203 1.5 -0.3 1.2 384123 7.2 4.3 11.5 381300 -0.1 -0.7 -0.8 384124 1.6 0.5 2.1 381401 -0.6 0.1 -0.5 384125 1.7 -0.1 1.5 381404 0.4 1.3 1.7 384126 3.8 5.8 9.6 381405 -0.3 -0.4 -0.7 382302 -3.3 -4.3 -7.6 381407 9.7 -6.3 3.4 383101 5.6 5.2 10.8 381408 0.3 -1.6 -1.2 383102 0.6 0.1 0.7 381409 0.6 -0.2 0.4 383103 0.5 5.7 6.2 381410 -0.1 0.3 0.2 383107 6.7 -5.8 0.9 381412 8.2 4.1 12.3 383108 -0.1 0.2 0.1 382101 2.2 1.1 3.3 383109 8.9 -8.8 0.1 382102 0.5 -0.7 -0.1 383110 -1.0 0.0 -1.0 382103 0.7 2.3 3.0 383201 -2.0 -3.1 -5.1 382104 0.1 -0.2 0.0 383204 -2.4 1.0 -1.4 382106 0.2 -1.3 -1.1 383205 0.9 -4.2 -3.3 382202 0.7 -1.3 -0.6 383206 0.1 -0.6 -0.5 382203 0.0 -0.3 -0.3 383301 0.2 2.1 2.3 382205 1.4 -0.3 1.1 383302 2.4 1.5 3.8 382206 2.9 5.4 8.3 383303 -0.2 -0.7 -0.9 382207 -0.2 0.1 0.0 383304 0.7 -1.9 -1.2 384201 0.4 -0.3 0.1 Total globalized 193.5 7.8 201.4 384202 0.6 0.7 1.3 384203 -0.8 0.2 -0.5 385001 -0.2 0.0 -0.2 385002 0.0 0.0 0.0 385004 -0.1 0.7 0.6 385005 -0.2 -0.1 -0.3 Total non-globalized 36.5 -2.0 34.5 Total sector 230.0 5.9 235.9

Source: prepared by the author, on the basis of data from the National Institute of Geography and Statistics (inegi), “Encuesta Industrial Mensual (eim). Clasificación Mexicana de Actividades y Productos (cmap), 205 clases de actividad económica”, 2013.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 145 metal sheet and aluminium” (381407) are symptomatic, improvements, labour attraction contributes less than because, like other classes and despite recording an 10% of the increase in labour productivity in the period increase in its labour productivity during the period, it studied (9.1%). This corroborates the inability of the experiences efficiency losses owing to labour layoffs leading industries to create new jobs to compensate (man-hours worked in class 381407 declined from for those destroyed in uncompetitive activities, as 19,458 in 1994 to 11,082 in 2008). noted above. In this regard, there is overwhelming evidence that In general, these findings contradict one of structural change (understood here as the shift of labour the main theoretical arguments for implementing a towards more efficient uses) contributes absolutely nothing trade liberalization strategy, namely the existence of to the trend of the indicator, both at the level of the 53 a creative-destruction process, in which jobs lost in classes and in respect of the two groups defined. Of the non-export activities, damaged by the public-policy total increase in labour productivity, the structural effect measures implemented, are offset by new job creation explains only 5.9 pesos of the 235.9 pesos in the sector as in the leading industries of the specialization pattern. a whole; it is responsible for 7.8 pesos of the 201.4 pesos A review of the data base used effectively shows that in the globalized group; and is actually negative in the the electromechanical sector, which encompasses the non-globalized group (see table 5). In fact, the structural pillar industries of Mexico’s manufacturing export effect is negative in 29 of the 53 classes considered. In model, shed 370,631 workers in net terms between the case of automotive assembly, which accounts for 1994 and 2008, in other words a reduction of 58,251 most of the sector’s technological and organizational man-hours worked.

V Productive specialization and dynamic competitiveness

In terms of export development, the foreign sales of To evaluate the trend of competitiveness in Mexico’s Mexico’s electromechanical sector grew vigorously electromechanical sector from a dynamic perspective, a by 380.2% between 1994 and 2008. As a counterpart, second database was created, in which 1,345 products however, the sector’s imports grew by a very similar identified as part of the sector in the hs classification amount (358.7%). In fact, in the 15 years of the were reclassified according to the Mexican Activities period indicated, the sector’s cumulative trade balance and Products Classification (cmap) maintained by the posted a deficit of US$ 76 billion (see figure 3). In National Institute of Statistics and Geography (inegi) 2008, for example, the sector’s exports amounted to (see annex). Based on this input, an extension of the US$ 169.6 billion: US$ 131.9 billion in the globalized methodology prepared for the magic programme by group and US$ 37.7 billion in the non-globalized group. the eclac Subregional Office in Mexico, was applied As a result of the liberalization dynamic, however, to 1,199 of these products.10 This exercise made it the sector as a whole reported an external deficit of possible to characterize the sector’s foreign sales by US$ 9.9 billion in that year, resulting from a trade their performance (dynamic or stalled) and based on surplus in the globalized group and a deficit in the non- the change in each product’s relative share of global globalized group. Deindustrialization in the latter group, demand in total goods trade (growing or declining) and the consequent gain of domestic-market shares by (eclac, 2006). The magic program thus establishes imports, thus combined with an organizational rationale a typology that classifies exports in terms of rising of the leading industries of the specialization pattern operating in global production networks, under which 10 As the necessary statistical information was not available for 146 most of the inputs included in the exported products products, the exercise was performed for 1,199 of the 1,345 products are imported. present in the database prepared.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 146 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 3 Cumulative exports, imports and trade balances by industry groups in the period 1994-2008 (US$ billion)

1 577 600 1 600 1 501 600

1 400

1 184 400 1 200 1 081 300

1 000

800

600 496 300

400 317 200

200 103 100 -179 100 -76 000 0 Exports Imports Trade balance -200 Globalized group Non-globalized group Sector total

Source: prepared by the author, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade).

stars (rs), waning stars (ws), lost opportunities (lo), total sales abroad in 2008.12 Of these products, 215, and regressions (r).11 representing 45.5% of total exports in that year, were The results of the exercise, which compared classified as rs —in other words goods with growing the years 1994 and 2008, are presented in table 6, trade in global markets in which Mexico increased its where the total value of exports for each typology relative share—. The statistical evidence thus reveals was calculated with respect to 2008. The sector total a sustainable competitiveness of the global production confirms the leadership of the electromechanical sector networks into which these leading industries of the within the Mexican export specialization pattern and its Mexican manufacturing sector are integrated, and an integration into global production networks through the export dynamism that is heavily concentrated in a small sale of dynamic goods, which, in the period studied, number of goods. increased their share in world merchandise trade. Of the In terms of the groups constructed, the exports of 1,199 goods considered, Mexico increased its international the non-globalized group display high levels of both market share in 710, representing 82.8% of the sector’s diversification and dynamic competitiveness. In 2008, despite being responsible for just 22.3% of the sector’s total exports, this group had foreign sales in 881 of 11 “Rising star” means that imports of the product in question increased in the United States market and that the country in question gained the 1,199 products registered. Of these sales, 78.8% a larger share of total imports of the product in the United States of the value corresponds to merchandise in which market. “Waning star” means that imports of the product in question these industries increased their global market share, decreased in the United States market, and that the country in question gained a larger share of the total imports of that product in the United and 38.5% of their value was obtained from “Rising States. “Lost opportunity” means that the imports of the product in stars” (rs). As table 6 shows, some of these relative question grew in the United States market but the country in question figures are similar in the case of the globalized group accounted for a smaller share of total imports of that product in the United States. “Withdrawn” means that the imports of the product in (318 products of the 1,199 registered, representing the question decreased in the United States and that the country in question obtained a smaller share of total imports of that product in the United States market (Cordero, 2010, p. 26). The exercise was based on the world market instead of being restricted to the United States, and the 12 Hereinafter the sector totals are obtained from the sum total of exports comparison of export values was made for 1994 and 2008. of the 1,199 products considered, so as to keep the exercise consistent.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 147

TABLE 6 Typology of products exported by industry groups in 2008 (Number and US$ million)

Groups Rising star Waning star Lost opportunity Regression Total

Globalized group Number of products 54 128 40 96 318 Value 61 917 400 47 432 600 12 067 700 8 750 500 130 168 200 Non-globalized group Number of products 161 367 100 253 881 Value 14 377 500 15 046 800 4 725 200 3 200 800 37 350 300 Sector total Number of products 215 495 140 349 1 199 Value 76 294 900 62 479 400 16 792 900 11 951 300 167 518 500

Source: prepared by the author, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade).

sum of rs (47.6%) and ws (36.4%) and 84% of the value as rs (19 products), 34.3% as ws (16 products), 6.9% of total exports of the group). This would suggest that as lo (eight products) and 8.4% as r (seven products) certain industries which are not integrated into global (see figure 4). production networks, but in some cases have a presence A final note concerns the characteristics of the on the domestic market, still have the organizational and methodology used and features that are often present technological capacities to be able to compete in the calculation of the indicators that exist to evaluate global domain. competitiveness. In the case of Mexico, the rapid trade- An analysis of the leading export products in 2008 liberalization process implemented by public policy confirms both the heavy concentration of the activity from the 1980s onwards fuelled a generalized increase in a few tasks and also the restricted aspect of export in national exports that outpaced the average expansion dynamism achieved. The 50 leading export products of international trade generally. This partly explains the represent 71.5% of the sector’s total exports; the top 10, increase in the global-market shares of the “leading” all relating to the manufacture of electrical-electronic and products in the Mexican specialization profile. Another computer equipment, or else related to the automotive trend in the global context reveals the results obtained: industry, account for 43.3%; and just three13 of the 1,1,99 dynamic goods, by definition, increase their share in products considered explain 28.4% of the sector’s sales international trade over time, so there is a bias in the abroad. Moreover, when these 50 products are shown methodology used that makes it more likely that the in a diagram on the basis of the established typologies, country being analysed will seem to evolve towards more the results do not differ greatly from those seen for competitive structures. In the case of the methodology the information from the entire database. As would developed by eclac, the established typologies are be expected given the leadership of those products in constructed by eclac trade in goods that are made from the Mexican manufacturing specialization pattern, the a whole series of components and inputs produced in figures are slightly higher in the rs category, which different parts of the world. As it was impossible to means an increase in global market shares in dynamic consider the domestic value-added content of the goods, goods during the period studied. In terms of the value of the methodology measures trade by the final price of foreign sales of these 50 products, 50.4% are classified the goods, imprecisely attributing capacities, capital investment, and hours worked to the trade shares. In the case of countries that have maquila-type productive 13 In the hs International Classification these goods are: 852810, Television receivers, including video monitors and video projectors; structures, such as Mexico, the exercises tend to overstate 870323, Other vehicles, with spark-ignition internal combustion the evolution of the export structure, as shown both by reciprocating piston engine, with cylinder capacity between 1,500 cc negative trade balances of the activity and the relative and 3,000 cc; and 852520, Transmission apparatus incorporating reception apparatus, etc. absence of spillover effects documented above.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 148 CEPAL REVIEW 114 • DECEMBER 2014

FIGURE 4 Typology of the 50 main products exported by the Mexican electromechanical sector, 2008 (Percentage differences in 2008 with respect to 1994)a

0.35

Lost opportunity Rising star 0.15

-0.05

-0.25

-0.45

Regression Waning star -0.65 Share of the product in world market

-0.85

-1.05 -24 -19 -14 -9 -4 1 6 11 16 Mexico’s share of the world marker for the product

Source: prepared by the author, on the basis of data from the United Nations Commodity Trade Statistics Database (comtrade). a In category 870422 the 1995 value was used because there were no data for 1994 in the information source.

VI Conclusions

The calculation of different indicators and statistical rationalized, in classes that are not engaged in global exercises, based on an extensive compilation of data production networks. The latter, in some cases maintain from different information sources, confirms a number the capacity to be competitive on the world stage, of trends observed in the structural evolution of the despite being immersed in a group that is undergoing Mexican electromechanical sector, which is a pillar obsolescence and deindustrialization. There is also of the nation’s export specialization profile. Firstly, overwhelming evidence of the absence of a significant efficiency improvements and supply have both become structural effect that would allow labour to migrate increasingly concentrated in a small number of firms that towards more productive uses, thereby partially negating are subsidiaries of transnational automotive assembly the orthodox theoretical argument used to defend the enterprises; and this has led to the breakdown of local capacity of the market to assign resources efficiently in value chains. Secondly, the leading industries of the the economy. export model have been unable to create jobs on a In terms of global competitiveness, the impossibility sustained basis or generate substantial technological and of obtaining a long and consistent statistical series that organizational spillover and diffusion effects. quantifies international trade in terms of the value-added The structural analysis also confirms the hypothesis present in the goods, is one of the factors making it that intrasectoral heterogeneity initially increases within harder to comprehensively evaluate the sector’s export the globalized activities following trade liberalization, performance. The results obtained with these constraints and then, once the productive apparatus has been provide evidence of sustainably dynamic activity of the

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 149 global production networks in which firms resident in moulded local capacities according to the competitive Mexico operate, as leaders of the model. Nonetheless, requirements of global production networks, resulting, the exports of industries that are not involved in global among other things, in a twofold specialization of the production networks are also highly diversified and leading industries in labour-intensive tasks (Deardoff, dynamic; and some eminently globalized classes of the 1979). Nonetheless, the incapacity of the strategy to electronics subsector display a phenomenon of “spurious” generate a significant structural change in terms of national competitiveness, characterized by growing foreign sales productive development calls the chosen guidelines into accompanied by an adverse trend in labour productivity. question. An alternative proposal, aimed at developing The role and results of the public policy implemented the strategic structural complementarities needed to in Mexico as from the 1980s, under orthodox theoretical increase the density and degrees of diversification in guidelines, thus seem to confirm the theses that several the productive system, could therefore be centred on authors have presented in introductory fashion. As claimed boosting the domestic market and, initially, on satisfying by Coe and others (2004), the measures established have the population’s basic needs.

ANNEX

TABLE A.1 Characteristics of the database on the productive structure of the Mexican electromechanical sector

Time Sector Units Variables Sources coverage coverage Mexican pesos at December Production, man- Monthly 13 subsectors The need to obtain lengthy and time-consistent 2003 prices. The data were hours worked and and annual. and 53 series that could reflect possible transformations deflated using the National its quotient, the Period classes of linked to structural change processes, meant Producer Price Index (inpp) labour- productivity 1994-2008. activitya that the only viable data source was the inegi of the manufacturing sector, indicator. 28 620 data Monthly Industrial Survey (eim) (inegi, 2013a) calculated by Banco de items in total. under the Mexican Activities and Products México (banxico) (2011). Classification (cmap).

Source: prepared by the author. a The globalized group contains 23 activity classes forming part of the following subsectors: 3841, “Automotive industry” (7 classes); 3823, “Manufacture and/or assembly of office, calculation, and information technology processing machines” (1 class); 3831, “Manufacture and/ or assembly of electrical machinery, equipment and accessories, including for electric power generation” (7 classes); 3832, “Manufacture and/or assembly of electronic equipment for radio, television, communications and medical use” (4 classes); 3833, “Manufacture and/or assembly of appliances and accessories of domestic use. Excludes electronic appliances” (4 classes). The non-globalized group includes 30 activity classes forming part of the following subsectors: 3811, “Smelting and moulding of metallic pieces, ferrous and nonferrous” (1 class); 3812 “Manufacture of metallic structures, tanks and industrial boilers. Including blacksmith work” (3 classes); 3813, “Manufacture and repair of metallic furniture” (1 class); 3814, “Manufacture of other metallic products. Excludes machinery and equipment” (8 classes); 3821, “Manufacture, repair and/or assembly of machinery and equipment for specific purposes, with or without an integrated electric motor. Includes agricultural machinery” (5 classes); 3822, “Manufacture, repair and/or assembly of machinery and equipment for general uses, with or without an integrated electric motor. Includes mountings” (5 classes); 3842, “Manufacture, repair and/or assembly of transport equipment and parts. Excludes automobiles and trucks” (3 classes); 3850, “Manufacture, repair and/or assembly of precision instruments and equipment. Includes surgical instruments. Excludes electronic instruments” (4 classes). Activity classes 382301, 383202, 384204 and 385006 were not considered because there was no information on them in the survey as from 2003.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ 150 CEPAL REVIEW 114 • DECEMBER 2014

TABLE A.2 Characteristics of the database on the external performance of the Mexican electromechanical sector

Time Units Variables Sector coverage Sources coverage U.S. dollars at - Exports from Mexico to the rest of the world Annual. 1 345 products United Nations Commodity current prices and by product and group. Period identified as part of Trade Statistics Database percentages. - Imports into Mexico from the rest of the 1994-2008 the sector under (comtrade). world by product and group. the Harmonized - Share of Mexican exports of each product in System (HS 92). the national total. 161 400 data items - Mexico’s share in the world market in total.a for the product. - Total world exports by product and group. - Share of each product in world exports. - Change in Mexico’s market share by product. - Change in the share of world exports by product. - magic typology by product.

Source: prepared by the author. a To estimate the value of the exports of activity classes regrouped in the cmap for the period 1994-2008, products manufactured by the electromechanical complex were differentiated under the HS 92 classification. As there are no official disaggregating equivalences between the Mexican Industrial Classification System and those commonly used for international trade, the 1,345 products at the six-digit level identified in HS 92 were reclassified according to the activity classes of the Mexican Activities and Products Classification (cmap).

Bibliography Álvarez, M. (2002), “Cambios en la industria automotriz frente a la Dawley, S. (2011), “Transnational corporations and local and regional globalización: El sector de autopartes en México”, Contaduría development”, Handbook of Local and Regional Development, y Administración, No. 206, Mexico City, National Autonomous A. Pike, A. Rodríguez-Pose and J.J. Tomaney (eds.), University of Mexico, July. London, Routledge. Arndt, S. (2001), “Offshore sourcing and production sharing in Deardoff, A. (1979), “Weak links in the chain of comparative preference areas”, Fragmentation. New Production Patterns advantage”, Journal of International Economics, vol. 9, No. 2, in the World Economy, S. Arndt and H. Kierzkowski (eds.), Elsevier, May. Oxford, Oxford University Press. Durán, J. and M. Álvarez (2008), “Indicadores de comercio exterior Arndt, S. and H. Kierzkowski (2001), “Introduction”, Fragmentation. y política comercial: Mediciones de posición y dinamismo New Production Patterns in the World Economy, S. Arndt comercial”, Project Document, No. 217 (LC/W.217), Santiago, and H. Kierzkowski (eds.), Oxford, Oxford University Press. Chile, Economic Commission for Latin America and the banxico (Banco de México) (2011), “Estadísticas en línea. Índice Caribbean (eclac). de precios al productor” [online] http://www.banxico.org. eclac (Economic Commission for Latin America and the Caribbean) mx/SieInternet/consultarDirectorioInternetAction.do?accion= (2007), “Progreso técnico y cambio estructural en América consultarDirectorioCuadros§or=20§orDescripcion= Latina”, Project Document, No. 136 (LC/W.136), Santiago, Chile. %CDndices%20de%20Precios%20Productor%20y%20de%20 (2006), “magic plus. Module to Analyse the Growth of Comercio%20Exterior&locale=es. International Commerce” [online] http://www.cepal.org/ magic/ Capdevielle, M. (2005), “Globalización, especialización y noticias/software/9/27199/magic%20Plus.pdf. heterogeneidad estructural en México”, Heterogeneidad Furtado, C. (1967), Development and Underdevelopment, University estructural, asimetrías tecnológicas y crecimiento en América of California Press. Latina, M. Cimoli (ed.), Santiago, Chile, Economic Commission Hernández, E. (2000), La competitividad industrial en México, for Latin America and the Caribbean (eclac)/Inter-American Mexico City, Plaza y Valdés. Development Bank (idb). inegi (National Institute of Statistics and Geography) (2013a), Coe, N. and others (2004), “‘Globalizing’ regional development: “Encuesta Industrial Mensual (eim). Clasificación Mexicana a global production networks perspective”, Transactions of de Actividades y Productos (cmap), 205 clases de actividad the Institute of British Geographers, vol. 29, No. 4, Wiley, económica” [online] http://dgcnesyp.inegi.org.mx/cgi-win/ December. bdieintsi.exe/ NIVZ101290009000020002#ARBOL. comtrade (United Nations Commodity Trade Statistics Database) (2013b), “Encuesta Industrial Anual. Clasificación Mexicana (2012) [online] http://comtrade.un.org/db/. de Actividades y Productos (cmap), 205 clases de actividad Cordero, M. (2010), “Indicadores de comercio exterior: Subregión norte económica” [online] http://dgcnesyp.inegi.org.mx/cgi-win/ de América Latina, 2010”, Working Paper (LC/MEX/L.995), bdieintsi.exe/NIVZ1012900110#ARBOL. Mexico City, Economic Commission for Latin America and (2010), “El sector automotriz en México 2010”, Estadísticas the Caribbean (eclac). Sectoriales series, No. 25, Mexico City.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ CEPAL REVIEW 114 • DECEMBER 2014 151

MacKinnon, D. (2012), “Beyond strategic coupling: reassessing the Unger, K. (2001), “La organización industrial, productividad y firm-region nexus in global production networks”,Journal estrategias empresariales en México”, Economía Mexicana. of Economic Geography, vol. 12, No. 1, Oxford, Oxford Nueva Época, vol. 10, No. 1, Mexico City, Economic Research University Press, January. and Teaching Centre, January-June. Ocampo, J. (2005), “La búsqueda de la eficiencia dinámica: Vázquez, R. (2013), “Determinants of structural heterogeneity in Dinámica estructural y crecimiento económico en los países Mexican manufacturing industry, 1994-2008”, cepal Review, en desarrollo”, Más allá de las reformas: Dinámica estructural No. 109 (LC/G.2556-P), Santiago, Chile, April. y vulnerabilidad macroeconómica, J. Ocampo (ed.), Bogota, (2012), “Heterogeneidad y cambio estructural en el D.C., eclac/Alfaomega. sector manufacturero mexicano”, Globalización y dinamismo Phelps, N.A. (2008), “Cluster or capture? Manufacturing foreign manufacturero. México y otros países emergentes, M.L. direct investment, external economies and agglomeration”, González and B. Olmedo (coords.), Mexico City, Economic Regional Studies, vol. 42, No. 4, Taylor and Francis, May. Research Institute, National Autonomous University of Mexico. Pinto, A. (1970), “Naturaleza e implicaciones de la ‘heterogeneidad Yamashita, N. (2008), “The impact of production fragmentation on estructural’ de la América Latina”, El Trimestre Económico, skill upgrading: new evidence from Japanese manufacturing”, vol. 37, No. 145, Mexico City, Fondo de Cultura Económica, Working Papers in Trade and Development, No. 2008/06, January-March. Canberra, The Australian National University. (1965), “Concentración del progreso técnico y de sus frutos Yeats, A. (2001), “Just how big is global production sharing?”, en el desarrollo de América Latina”, El Trimestre Económico, Fragmentation: New Production Patterns in the World vol. 32, No. 125, Mexico City, Fondo de Cultura Económica, Economy, S. Arndt and H. Kierzkowski (eds.), Oxford, Oxford January-March. University Press. Puyana, A. and J. Romero (2006), “Hacia una evaluación de los Yeung, H. (2009), “Transnational corporations, global production efectos multiplicadores de la actividad maquiladora”, Estudios networks and urban and regional development: a geographer’s Sociológicos, vol. 24, No. 1, Mexico City, El Colegio de perspective on multinational enterprises and the global México, January-April. economy”, Growth and Change, vol. 40, No. 2, Wiley, June.

GLOBAL INTEGRATION, DISARTICULATION AND COMPETITIVENESS IN MEXICO’S ELECTROMECHANICAL SECTOR: A STRUCTURAL ANALYSIS • RAÚL VÁZQUEZ LÓPEZ

Technological capacity-building in unstable settings: Manufacturing firms in Argentina and Brazil

Anabel Marín, Lilia Stubrin and María Amelia Gibbons

ABSTRACT From the 1970s onward, the macroeconomic context in Argentina and Brazil was characterized by drastic economic changes and instability. Numerous studies have documented the generally negative effect of this environment on the innovation capacities of the manufacturing sector. This paper, however, analyses the possible emergence of new innovation capacities in the period, bringing two important phenomena to light. First, a quite substantial number of firms, even in unstable settings, redoubled their innovation efforts. Second, these firms are mainly found in a small group of sectors associated with the countries’ static advantages or in sectors favoured by specific sectoral regimes. The findings, although exploratory, are a contribution to the debate on the development of innovative capacities in unstable macroeconomic contexts and the ability of sectors associated with the two countries’ static advantages to generate spaces of innovation and value creation.

KEYWORDS Industry, manufactures, economic conditions, technological innovations, research and development, statistical data, Argentina, Brazil

JEL CLASSIFICATION O14, O25, O32

AUTHORS Anabel Marín is a researcher with the National Council for Scientific and Technological Research (conicet), the Research Centre for Industrial Transformation (cenit) and Tres de Febrero National University (untref ), Argentina. [email protected] Lilia Stubrin is a post-doctoral fellow at the Research Centre for Industrial Transformation (cenit), Argentina. [email protected] María Amelia Gibbons is a research assistant with the Research Centre for Industrial Transformation (cenit) and a teaching assistant at the University of San Andres, Argentina. [email protected] 154 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

This paper studies the experience of Argentine and These fluctuations and sudden shifts in the political Brazilian manufacturing firms with the accumulation and economic context are at the centre of the present study of innovative capacities over the past two decades. on technological capacity-building in the region, which The time frame chosen coincides with a context of takes manufacturing industry in Brazil and Argentina as economic instability and changing economic policies in its case studies.2 Many earlier works have characterized both countries. This is unusual in studies of innovative the innovation pattern of firms in the Latin American capacity-building in emerging countries, which have countries over the past few decades, including Katz tended to analyse long periods of continuous accumulation (2001 and 2007), Cimoli and Katz (2003), Chudnovsky, covering two, three or more decades.1 Such studies reveal López and Pupato (2006), Baer (1970 and 1972), Teitel long-term continuity on two levels. First, they usually and Thoumi (1987) and Erber, Guimarães y Tavares show capacity development at the microeconomic level Araújo Jr. (1974). Most of these studies have analysed the progressing without complications through successive generally negative repercussions of the economic reforms stages of ever “deeper” innovation activities. Second, of the 1980s and 1990s for the innovation capacity of in the relatively few Asian countries where most of the the manufacturing sector.3 However, this study adopts knowledge about the long-term technological behaviour a rather different perspective. The question needs to of industrial firms in emerging countries has been be asked as to whether there has been any process of generated, institutional and political contexts evolved new innovation capacity-building in the manufacturing continuously and were fairly stable. Although there were sector during this period of drastic economic change and changes in the main policy emphases and some political macroeconomic instability (when the prevailing effect and economic shocks, crises were not far-reaching on firms’ innovation capacities has been negative), what or damaging enough to destroy and disperse existing the sectoral ecology of this process is, and in what types innovation capacities. Consequently, the main lines of of firms these capacities have arisen. The focus of the technological development were rarely disrupted, let present research is on the new clusters of innovative alone cut short or reversed. firms created following the structural transformations The Latin American experience has been very in these countries during the 1990s. It should be pointed different. Following a period of cumulative firm- out that the approach is dynamic and relative, not static level capacity-building in the context of the relatively or absolute. The focus is not on individual cases of stable regimes and policies of the import substitution successful firms at a given point in time but rather on industrialization (isi) period between the 1950s and 1970s, the way the technological efforts and outcomes of the instability and fluctuations in the business cycle became main clusters of innovating firms have evolved. recurrent. Not only have most of the region’s countries been The empirical analysis is based on evidence from affected by acute economic and financial crises, but they innovation surveys. In Argentina, the surveys available have implemented far-reaching changes in the orientation cover the period from the early 1990s to the mid-2000s, of economic policy in the last few decades. Between the while in Brazil they cover a shorter period in the 2000s. 1980s and 1990s, trade and finance were liberalized, State The analysis centres on firms that were in the vanguard involvement in the economy was reduced, currencies of innovation in the countries and periods analysed. appreciated and foreign investment was favoured. These firms were identified by taking those that claimed Thereafter, the State gradually recovered its influence to have introduced a product or process innovation (or on the economy and neoliberal policies began to retreat.

2 A forerunner of this study is Marín and Bell (2012), centring on 1 For example, Amsden (1989) in the Republic of Korea between Argentina in 1992-2001. the 1960s and 1980s; Kim (1997) on accumulation processes in the 3 Although some innovation activities were carried out with positive Republic of Korea; Hobday (1995) in Singapore, Taiwan Province of results, particularly productivity increases and a narrowing of the China and the Republic of Korea in the 20 to 30 years after the 1970s; efficiency gap with leading countries as a result of technology investment and Mathews and Cho (2000) in their study of the development of the in several areas of the economy, especially industries associated with semiconductor industry in Asian countries. the exploitation and processing of natural resources.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 155 both) that was new to the world economy at the start of raised, the intention being rather to bring new empirical the period studied. This innovation outcome indicator, evidence to the debate about the generation of innovation while subjective, does make it possible to identify the advantages in the region’s countries. The information most significant end of the innovative firms distribution. analysed does not cover the past few years, since innovation The evidence analysed reveals some striking (or surveys are unavailable from 2005 onward. However, not wholly foreseeable) facts. First, new clusters of that does not take away from the main conclusions innovative firms seem to have been emerging despite of this study, since to a large extent the phenomena the crises and far-reaching changes in economic policy identified have longer-term repercussions. Lastly, it orientation that have characterized these countries in should be stressed that the purpose of this paper is to recent decades. This evidence, although preliminary, present Argentina and Brazil as two case studies, but goes against a pessimistic view in the region that has not to draw a comparison between them. The difficulty mainly identified negative results. It is interesting to of comparing lies in the fact that different samples of note, however, that these clusters of innovators are dissimilar firms are taken for different periods, and that found in a small number of sectors protected by static the indicator used to identify innovative firms, being location advantages (such as traditional sectors or those subjective, could be influenced by national biases in associated directly or indirectly with natural resources) the way the information employed in it is interpreted and indeed some sectors favoured by public policies, and understood. such as the automotive sector, which benefits from a The article is structured as follows. Section II special protection regime in both Brazil and Argentina. briefly describes the context of instability in which Lastly, and once again challenging pessimistic notions manufacturing firms in Argentina and Brazil have about the innovation intensity and potential dynamism operated since the relatively stable period of protection of sectors protected by static advantages (such as during isi. Section III lays out the data and methodology. natural resource-linked and traditional sectors), it can Section IV presents the first set of findings, identifying be seen that firms in these sectors have been making the sectors to which the most innovative firms belong. substantial, above-average efforts to improve their Section V concentrates on characterizing particularly innovation and technology performance, albeit from very innovative firms, identifying their origins and history low levels. and their patterns of technological behaviour. Section This research is wholly exploratory. No attempt VI presents the implications of all this and provides is made to give definitive answers to all the questions some reflections based on the findings.

II The context: from protection to exposure in a setting of profound instability

1. The beginnings of industry: the diversified and fairly sophisticated manufacturing sector protectionist stage that had begun to act as a driver of their economies. However, industrialization in Argentina and Brazil The industrial and technological development of was heavily dependent on imported inputs and capital manufacturing in Argentina and Brazil began almost goods from the start. This characteristic, combined with spontaneously in the late nineteenth century, when the weak export growth, led to recurrent balance-of-payments agricultural export model was dominant in both countries. deficits which fed through to the main macroeconomic In the 1930s, however, the State began to actively promote variables, creating regular crises, instability and inflation. industrialization by implementing policies to protect Many authors saw the inability of the manufacturing the domestic market (i.e., tariff barriers), creating State sector to generate exportable growth, reduce import institutions (lending institutions and development banks) dependence and secure steady as opposed to stop-start and providing infrastructure. As a result of these policies, growth as being due particularly to certain limitations and four decades later Argentina and Brazil had developed a imbalances that characterized technological learning and

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 156 CEPAL REVIEW 114 • DECEMBER 2014 industrial capacity-building (Katz, 1972, 1987, 2001 and In the 1990s, the liberalization and deregulation 2007; Katz and Kosacoff, 1989 and 2000; Dahlman, 1984; process initiated in the mid-1970s was carried further, Dahlman and Fonseca, 1987; United Nations, 1969). It although some novel elements also emerged: the has been argued that there was substantial growth in the privatization of most public-sector enterprises, price manufacturing sector during the period, accompanied by stabilization4 and currency appreciation. These elements steady increases in aggregate productivity and significant imposed a new competition environment on industry. technological learning, particularly in some low- and Studies carried out in the late 1990s and early 2000s medium-technology sectors (Teitel and Thoumi, 1987). identified the following trends as a consequence of the However, there were two major constraints. First, transformations wrought in the 1990s: excessive vertical integration and limited specialization (i) A slowdown in industrial growth: in Argentina, in engineering, combined with an abundant supply of the contribution of the industrial sector to gross skilled labour, resulted in an excessive domestic focus domestic product (gdp) fell from 19% in 1990 to on adapting and incrementally improving technologies at 16% in 2000, while in Brazil industry grew at a the expense of investment in incorporated technological slower rate even as it maintained its gdp share at change or research and development (r&d) aimed at a similar level throughout the 1990s (Barros de significantly reducing costs and developing more “radical” Castro, 2003; Ferraz, Kupfer and Iootty, 2004). innovations and new products (Teitel, 1981; López, 1996). (ii) Changes at the sectoral level: some industries Indeed, the evidence is that spending on r&d, royalties, shrank,5 some new ones were added to the industry patents and franchises was low by international standards mix and others disappeared. Industrial and natural for the time (Katz, 1972). Second, inward technology resource-based commodities increased their presence transfer within sectors was very limited (Dahlman and (Cimoli and Katz, 2003; Ferraz, Kupfer and Frischtak, 1990). In each sector, a small set of firms Iootty, 2004). that had succeeded in attaining global standards of (iii) Increased heterogeneity and concentration: some productivity through innovation efforts in engineering, firms, mostly large ones and subsidiaries of quality, design and organization usually coexisted with multinationals, substantially raised their productivity a mass of firms characterized by low productivity, little and matched international best practices by investing innovation effort, obsolete equipment and products that in equipment and implementing organizational were of low quality by international standards. It has been changes (introducing process automation, for argued that the lack of external competition was a decisive example). However, a great many firms, particularly factor in discouraging technological modernization and small and medium-sized ones (smes), were unable innovation across much of the production spectrum. to implement these changes and disappeared, while others simply hung on, improving their productivity 2. Structural reforms: destruction essentially by shedding employees (Kosacoff, 1996, and resurgence 2000a and 2000b; Ferraz, Kupfer and Iootty, 2004; Katz and Bercovich, 1993). The progressive industrialization that took place, with (iv) The closing of the productivity gap: in many sectors, fluctuations, throughout the twentieth century came modernization was carried out by rationalizing costs to a sudden halt in the mid-1970s as economies were and acquiring technologies from abroad. Tie-ups opened up and regulations and subsidies protecting the with international suppliers, licensing and online industrial sector were removed (Katz and Kosacoff, technical assistance became the preferred channels. 2000). The economic instability and debt crisis of the At the same time, local equipment suppliers, 1980s further entrenched the process of economic and engineering firms and r&d laboratories lost ground industrial stagnation. In consequence of these contextual (Katz, 2001; Ferraz, Kupfer and Haguenauer, 1996; and policy changes, manufacturing in Argentina and Ferraz, Kupfer and Iootty, 2004). Brazil underwent the deepest crisis in the history of the (v) Transnationalization: Argentina and Brazil industrial sector in the late 1970s and early 1980s. This became the main destinations for foreign direct crisis manifested itself, among other things, in the exit of firms from the market, a sharp fall-off in investment 4 Price stabilization following enactment of the 1991 Convertibility and the increasing obsolescence of capital equipment Act in Argentina and the 1994 Real Plan in Brazil. (see, among others, Katz, 2001; Kosacoff, 1996; Ferraz, 5 In Argentina, the metallurgy and chemical industries, the two most Kupfer and Serrano, 1999; Baer, 2001). dynamic during the isi period, contracted in the 1990s.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 157

investment (fdi) in Latin America. In Argentina, the form of takeovers of existing firms rather than the foreign share of total sales by the 200 largest an increase in gross fixed capital. manufacturing firms increased from 43% in 1994 These characteristics arose in a context of great to 69% in 1998 (Kulfas, Porta and Ramos, 2002). macroeconomic instability: periods of stabilization In Brazil, multinationals increased their share of and growth followed by others with negative growth sales by the country’s 300 largest firms from 14.8% rates, shocks from international crises and external to 36.4% (Rocha and Kupfer, 2002). Incoming vulnerability. The main focus of the present analysis capital was used mainly to acquire existing assets is on the generation of innovative capacities in via privatization, mergers and acquisitions. Thus, the manufacturing sector from the 1990s to the the incorporation of foreign capital tended to take mid-2000s.

III Data and methodology

1. Data are particularly inappropriate in contexts such as those of Argentina and Brazil, where firms rarely The data used are from innovation surveys carried out use patents to protect their new knowledge. This is among manufacturing firms in Argentina and Brazil. mainly for two reasons: (i) firms are usually adopters In the case of Argentina, surveys covering the periods of technologies at or close to the frontier; (ii) there 1992-1996 and 2002-2004 were used, while for Brazil is a sectoral bias in the region towards process surveys from 2000 and 2005 were analysed. Because industries, where there is usually less of a tendency the samples of firms altered between one survey and to patent. In Argentina, for example, according to the next,6 the data are based on a subsample of firms data from the National Institute of Statistics and featuring in every survey considered in each country. Censuses (indec), the Secretariat of Science and This subsample contains 608 firms in Argentina and Technology (secyt) and the Economic Commission 3,890 in Brazil. for Latin America and the Caribbean (eclac), between 1998 and 2001 just 10% of innovators 2. Methodology: selecting innovative firms patented (Chudnovsky, López and Pupato, 2006). In this study, the concern is to capture the general The study aims to identify firms in the vanguard of characteristics of all innovative firms, not just those innovation activity in the Argentine and Brazilian that patent, meaning that indicators of this type are manufacturing sectors over recent decades, i.e., the not wholly suitable. subgroup of the 608 firms surveyed in Argentina and • Indicators based on r&d reflect the efforts made the 3,890 in Brazil that can in some way be categorized by firms to create knowledge, but do not show how as “particularly innovative”. It was not considered effective this activity is and, more importantly in appropriate to employ for this purpose either of the this context, tend to reflect differences in r&d two indicators most commonly used to measure firms’ intensity between industries rather than specific innovation capacity, namely patents and r&d spending. differences in innovation intensity between firms. • Results indicators based on patents usually have Furthermore, these indicators tend to underestimate limitations, as they reflect invention rather than innovation activities closely related to production innovation, and because they have a strong sectoral and information processing (Patel, 2000). bias (Scherer, 1983; Harabi, 1995; Levin and others, As innovation surveys have spread in many countries, 1987). Furthermore, indicators based on patents different measures have begun to be used in the past 15 years to evaluate the innovativeness of firms. The most popular, when the aim is to measure outcomes 6 In Argentina, the 1992-1996 innovation survey covered 2,430 firms and the 2002-2004 survey 1,690. In Brazil, the 2000 and 2005 innovation rather than efforts, has been the percentage of sales surveys covered 10,328 and 12,172 firms, respectively. accounted for by innovative products. In principle, this

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 158 CEPAL REVIEW 114 • DECEMBER 2014 is a good direct measure of innovation outcomes, but To avoid confusion, the term “significantly innovative it discriminates against process innovation (which is firms” will be used in the rest of the analysis for firms particularly important in Argentina and Brazil). This is replying that they had introduced product or process why a different measure was chosen for the present study. innovations or both that were new to the global economy. Following the standard distinctions of the Oslo Manual The rest will simply be termed “non-innovative firms”, (oecd, 1997), innovation surveys ask manufacturing even though the latter group includes firms claiming to firms about the degree of novelty entailed in innovations have introduced innovations that were new to the firm or to products or processes, or both, introduced by them the country. In the samples analysed, 68 Argentine firms into the market during the survey period. When firms are (11%) and 167 Brazilian firms (4.3%) were found to be asked about their innovation performance, four possible significantly innovative. Of these, most innovated only answers are available: in products (60% in Brazil and 67% in Argentina). A (i) no product (or process) innovations were introduced; larger share of firms in Brazil (28.1%) than in Argentina (ii) product (or process) innovations new to the firm (7.4%) introduced only process innovations into the were introduced; market. Those introducing both product and process (iii) product (or process) innovations new to the local innovations amounted to 12.6% of the total in Brazil economy were introduced; or and 25% in Argentina (see table 1). (iv) product (or process) innovations new to the world It should be pointed out that the subjective character economy were introduced.7 of the indicators used means that the findings for Argentina Firms answering yes to the last of these options in the and Brazil ought not to be compared, as they could reflect first innovation survey carried out in each country were national biases in the way the question is interpreted. taken to be the most innovative. An obvious constraint is that the answers are subjective. Answers by firms TABLE 1 claiming to have introduced this category of innovation Types of innovative firms in may not be a very accurate reflection of what innovations Argentina and Brazil actually are “new to the world market”. However, this (Numbers and percentages) kind of accuracy is not the main issue here, as the main concern is with the relative innovation capacity of firms Type of firm Argentina Brazil within each country, rather than with the identification Non-innovative 540 (88.8) 3 723 (95.7) of global leaders. The firms surveyed have the option of Significantly innovative 68 (11.2) 167 (4.3) Process innovation only 5 47 selecting less novel forms of innovation, and it would Product innovation only 46 99 seem that, at least in relative terms, this category does Product and process innovation 17 21 adequately capture the most significant end of the Total 608 (100) 3 890 (100) innovative firms distribution. Source: prepared by the authors, on the basis of the Survey on Technological Behaviour of Argentine Industrial Firms (1992-1996) and the National Survey of Firms on Innovation, Research and Development and Information and Communication Technologies 7 See the latest version of the Oslo Manual (oecd, 2005, pp. 46-47) (2002-2004) in Argentina, and the Survey on Technological for an explanation of these categories. Innovation (pintec) (2000 and 2005) in Brazil.

IV Analysis I: sectoral distribution of significantly innovative firms

This section studies the sectoral ecology of firms identified classified into two groups using the three-digit International as significantly innovative. In particular, the aim is to Standard Industrial Classification of All Economic establish whether there is a set of sectors where these firms Activities (isic): sectors including at least one significantly cluster, as this will reveal areas of dynamic comparative innovative firm, and sectors with no such firms. The advantages. With this in view, manufacturing sectors were latter were discarded for the purposes of the analysis.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 159

There are 31 sectors with at least one significantly with clusters of significantly innovative firms. Clustered innovative firm in Argentina and 70 in Brazil. It should innovative firms represent 18% of all firms in these sectors be stressed, however, that the distribution of significantly in Argentina and 7% in Brazil (see the fourth column innovative firms between sectors is far from homogeneous. in tables 2 and 3), i.e., 82% and 93% of firms in sectors There are sectors with many more significantly innovative with clusters of innovative firms are not significantly firms than others (see column 3 of tables 2 and 3). To innovative in Argentina and Brazil, respectively. This identify the sectors where significantly innovative firms suggests that the selection in this study encompasses cluster, the sectoral distribution of these firms was a fairly small group at the upper end of the corporate considered and sectors were classified into two groups: innovativeness distribution. (i) those where the concentration of firms is equal to In terms of sectoral ecology, one important or greater than the median of the distribution (“sectors characteristic to highlight is that four of the five sectors with clusters of significantly innovative firms”), and with clustered innovative firms in Argentina and four (ii) those where the concentration of firms is below the of the seven with the same characteristics in Brazil are median of the distribution (“sectors without clusters of associated via input-output links with natural resource- significantly innovative firms”).8 From here onward, based industries9 (see table 4). Argentina accounts for the firms in the first group of sectors will be referred to as bulk of innovative firms clustered in a group of sectors “clustered innovative firms” and those in the second directly and indirectly linked to natural resources. The group of sectors as “isolated innovative firms”. sectors directly linked to the exploitation and processing Tables 2 and 3 show that clustered innovative of natural resources are those that process agricultural firms are concentrated in a small set of sectors within and livestock products, while the sectors indirectly linked manufacturing industry: five sectors in Argentina and to the exploitation and processing of natural resources seven in Brazil. These sectors account for 42% and 31% are those supplying inputs for agricultural production of firms of this type in Argentina and Brazil, respectively. (fertilizer and machinery). These sectors involve the Some further observations should be made about type of production activity for which Argentina has the intrasectoral composition of the group of sectors a historical competitive advantage. It should also

8 The list of sectors without clusters of significantly innovative firms is available from the authors on request. 9 See annex for the methodology.

TABLE 2 Distribution of significantly innovative firms between sectors in Argentina (Three-digit isic categories)

Significantly innovative firms Significantly innovative firms in Sector in the sector as a share of all the sector as a share of all firms significantly innovative firms in the sector

(2) (3) (4) (1) Firms (percentages) (percentages) (number)

Clustered innovative firms 29 42 18 Manufacture of pesticides and other 9 13 17 agrochemical products Production, processing and preservation of meat, fish, 7 10 18 fruit, vegetables, oils and fats Manufacture of agricultural and forestry machinery 5 7 45 Manufacture of other food products 4 6 13 Manufacture of parts and accessories for 4 6 24 motor vehicles and their engines

Isolated innovative firms 39 58 9 Total 68 100 11

Source: prepared by the authors, on the basis of the Survey on Technological Behaviour of Argentine Industrial Firms (1992-1996). isic: International Standard Industrial Classification of All Economic Activities.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 160 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 3 Distribution of significantly innovative firms between sectors in Brazil (Three-digit isic categories)

Significantly innovative Significantly innovative firms in the sector as a firms in the sector as a Firms Sector share of all significantly share of all firms in (number) innovative firms the sector (percentages) (percentages)

(1) (2) (3) (4)

Clustered innovative firms 50 31 7 Manufacture of pharmaceuticals and medicinal 12 7 13 chemicals Manufacture of footwear and parts of footwear 8 5 4 Manufacture of parts and accessories for motor 7 4 6 vehicles and their engines Manufacture of basic chemicals, fertilizers, nitrogen 6 4 17 compounds, plastics in primary forms and synthetic rubber Manufacture of agrochemical products 6 4 24 Manufacture of pumps, compressors, taps and valves 6 4 12 Manufacture of plastic products 5 3 3 Isolated innovative firms 97 69 3 Total 167 100 4

Source: prepared by the authors, on the basis of the Survey on Technological Innovation (pintec) (2000 and 2005) in Brazil. isic: International Standard Industrial Classification of All Economic Activities.

be emphasized that a large part of the rationale for chemicals” and “manufacture of pumps, compressors, the isi regime was to shift production specialization taps and valves”. It is important to clarify that while away from sectors of this type, in part because they the sectors just mentioned are linked via input-output were regarded as “low-technology” sectors that did relationships to activities based on natural resource little to further the incorporation of more important processing (see annex), it would be a mistake to think innovative activities within the economy (especially that the whole set of firms belonging to these sectors are in the case of the industries most directly linked to necessarily related. For example, the “manufacture of natural resources). pharmaceuticals and medicinal chemicals” sector may In Brazil, sectors with clustered significantly include both innovative firms supplying inputs to natural innovative firms are more diversified between sectors resource-based activities and others not connected to connected and unconnected to activities based on natural resources (such as pharmaceutical firms working the exploitation and processing of natural resources. in the area of human health). Nonetheless, more than half of all clustered innovative Other sectors in which innovative firms cluster firms (60%) are involved with natural resource-based but which are not connected to natural resources are activities, typically as suppliers. Particularly prominent traditional ones such as “manufacture of footwear among sectors linked to natural resources as suppliers are and parts of footwear” and “manufacture of plastic those involved in the “manufacture of basic chemicals, products”. Clustered innovative firms that are not fertilizers, nitrogen compounds, plastics in primary forms linked to natural resources also appear in Argentina and synthetic rubber”, “manufacture of agrochemical and Brazil in sectors connected with automotive products”, “manufacture of pharmaceuticals and medicinal production (sectors 343 and 344). These reflect what

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 161

TABLE 4 Classification of sectors with clusters of significantly innovative firms in Argentina and Brazil

Sectors linked to natural resource-based industries Sectors not linked to natural resource-based industries Directly Indirectly

Argentina - Production, processing and - Manufacture of pesticides - Manufacture of bodies preserving of meat, fish, fruit, and other agrochemical (coachwork) for motor vehicles; vegetables, oils and fats (151) products (242) manufacture of trailers and - Manufacture of other food - Manufacture of agricultural and semi-trailers (343) products (including bread, forestry machinery (292) biscuits, sugar, cocoa, tea and maté, among others) (154)

Brazil - Manufacture of basic chemicals, - Manufacture of footwear and fertilizers, nitrogen compounds, parts of footwear (193) plastics in primary forms and - Manufacture of parts and synthetic rubber (241) accessories for motor vehicles - Manufacture of agrochemical and their engines (344) products (242) - Manufacture of plastic - Manufacture of pharmaceuticals products (252) and medicinal chemicals (245) - Manufacture of pumps, compressors, taps and valves (291)

Source: prepared by the authors, on the basis of input-output matrices for Argentina and Brazil.

remains of the flourishing automotive metallurgy Argentina (50%) and the “service activities incidental industry that was nurtured during the isi period and to oil and gas extraction excluding surveying” sector that since the early 1990s has been at the centre of a in Brazil (33%). system of sectoral policies which includes a special In summary, it transpires that firms identified as trading regime between the two countries and incentives significantly innovative in the countries analysed are for producers. clustered in a small group of sectors. In Argentina, four Sectors with clusters of innovative firms were deemed of the five sectors concerned are directly or indirectly more innovative, since firms of this type represent a larger linked to natural resource exploitation and processing percentage of all firms in these sectors than in sectors activities. There is more variability in the case of Brazil, without such clusters (see the fourth column of tables 3 with none of the sectors being directly connected to natural and 4). It must be stressed, however, that among the latter resources as customers, but there are some sectors that are a number of sectors where significantly innovative are clearly associated with natural resources as suppliers firms represent a larger proportion of the total than in and, strikingly, some traditional sectors too. Featuring the group of sectors where significantly innovative in both countries is the automotive sector, which has a firms cluster. However, these sectors contain very few special protection regime. Since both natural resource- firms, and they appear rather to be isolated instances linked activities (particularly those identified as having of innovation than areas of sectoral clustering. Some a direct link) and traditional ones are regarded as low- sectors with these characteristics are the “processing innovation and low-value added activities, this result, of nuclear fuel” sector in Argentina and Brazil (100%), taken by itself, is surprising. The next section goes into the “manufacture of insulated wire and cable” sector in this in depth.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 162 CEPAL REVIEW 114 • DECEMBER 2014

V Analysis II: characterization of clustered innovative firms

This section looks more closely at the characteristics of independent firms) and a larger share of subsidiaries of clustered innovative firms. The first aim is to understand foreign companies and local business groups (accounting the structural characteristics of these firms, such as the for 17% and 11% of the manufacturing total, respectively). origin of their capital, their size and their patterns of As regards size, clustered innovative firms are larger innovation behaviour. To this end, clustered innovative than the mean for the manufacturing sector (see table firms are compared with other groups of firms such as 6), with 48% of clustered innovative firms having over isolated innovators, non-innovators, firms in “high- 200 employees, as against just 39% of the whole set of technology” industries,10 the subsidiaries of foreign manufacturing firms. companies and the whole manufacturing establishment In terms of innovation behaviour, in the early 1990s in general. This is followed by an exploration of the clustered innovative firms in Argentina as a group made way firms’ innovation behaviour changed over the greater efforts in capital investment (1.33%) than in r&d period studied, with their technological behaviour being (0.10%). The pattern observed for firms of this type was evaluated using two types of innovation effort indicators: also seen for the other types of firms considered (see (i) the intensity of firms’r&d spending (r&d spending categories 2 to 6 of table 7). However, the difference as a share of total sales), and between the intensity of spending on capital goods (ii) the intensity of investment in capital goods for for innovation and spending on r&d was smaller for innovation (spending on capital goods as a share clustered innovative firms (taken all together) than for of total sales). the other groups of firms. Another interesting point to The evidence is presented first for Argentina and come out of the analysis is that clustered innovative then for Brazil. It is important to clarify that the data firms, together with those in high-technology sectors, on firms’ innovation behaviour in the two countries are are the ones that invested most in r&d as a proportion not comparable, as they cover different years and were of sales, although it might be noted that levels of r&d collected from different samples. The greatest difference spending were generally quite meagre. is that in Brazil the averages are calculated with reference When innovation behaviour is analysed separately to a sample of innovative firms (firms replying that they between the different groups of clustered firms, there have not innovated during the period are excluded), while are found to be substantial differences between them in Argentina the averages are calculated from the total that are worth highlighting. The innovation activity of sample of firms interviewed, innovative or otherwise. clustered firms linked to natural resources as customers was mainly based on investment in new technologies 1. Clustered innovative firms in Argentina already incorporated into capital goods. Internal r&d spending, on the other hand, ranked among the lowest, The majority of clustered innovative firms connected to along with that of non-innovative firms. Conversely, natural resources in Argentina11 are independent (52%), clustered firms indirectly linked to natural resources, while smaller proportions are subsidiaries of international along with high-technology firms, were among the companies (31%) and local business groups (17%) (see types of firms investing most in r&d in the early 1990s. table 5). The composition of this set of firms differs They spent 0.17% of sales on r&d activities, whereas from that of the whole set of Argentine manufacturing manufacturing firms as a whole invested 0.06% of sales firms in that it includes a smaller share of independent in the period. firms (71% of the manufacturing sector is composed of As for innovative firms clustered in industries linked directly to natural resources, the pattern almost exactly matches the observations made in earlier 10 Using the high-technology firms classification of the Organization for Economic Cooperation and Development (oecd). studies of innovation activity in the 1990s. Innovation 11 In Argentina, the focus is essentially on firms clustered in natural largely consisted in investments that were intensive in resource-linked sectors, as they account for the bulk of the total.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 163

TABLE 5 Firms in Argentina by ownership type (Numbers and percentages)

Innovative firms clustered in natural Isolated Non-innovative firms Manufacturing resource-linked innovative firms in all industries sector total industries Local business groups 5 7 57 69 (17) (18) (11) (11) Independent local firms 15 24 395 433 (52) (62) (73) (71) Subsidiaries of multinationals 9 8 88 105 (31) (21) (16) (17) Total 29 (100) 39 (100) 540 (100) 608 (100)

Source: prepared by the authors, on the basis of the Survey on Technological Behaviour of Argentine Industrial Firms (1992-1996) and the National Survey of Firms on Innovation, Research and Development and Information and Communication Technologies (2002-2004).

TABLE 6 Firms in Argentina by size (Numbers and percentages)

Innovative firms clustered in natural Isolated innovative Non-innovative firms Manufacturing resource-linked firms in all industries sector total industries smes 15 27 327 369 (52) (69) (60) (61) Large firms 14 12 213 239 (48) (31) (40) (39) Total 29 (100) 39 (100) 540 (100) 608 (100)

Source: prepared by the authors, on the basis of the Survey on Technological Behaviour of Argentine Industrial Firms (1992-1996) and the National Survey of Firms on Innovation, Research and Development and Information and Communication Technologies (2002-2004). Note: small and medium-sized enterprises (smes) are those with 200 employees or less and large enterprises those with at least 200 employees.

TABLE 7 Argentina: innovation behaviour in the manufacturing sector, 1992

r&d intensity Capital investment

Types of firms Average values per firm (percentages)

1. Innovative firms clustered in sectors: 0.10 1.33 - directly linked to natural resources 0.01 1.49 - indirectly linked to natural resources 0.17 1.23 - not linked to natural resources 0.12 -10.16 2. All other innovative firms 0.09 1.97 3. Firms in “high-technology” industries 0.20 1.57 4. Subsidiaries of foreign multinationals 0.06 3.17 5. Non-innovative firms 0.01 1.47 6. All firms 0.06 1.81

Source: prepared by the authors, on the basis of the Survey on Technological Behaviour of Argentine Industrial Firms (1992-1996) and the National Survey of Firms on Innovation, Research and Development and Information and Communication Technologies (2002-2004). r&d: research and development.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 164 CEPAL REVIEW 114 • DECEMBER 2014 technologies already incorporated into capital goods, the start of the period (such as firms in high-technology and production capacity was substantially enhanced as sectors). Clustered firms not linked to natural resources, a result. However, there was virtually no sign of r&d conversely, were the only ones to show negative rates within firms themselves, and innovative knowledge of increase in r&d spending. creation capacity was hardly in evidence. In the case In 1992-2004, investment in capital goods for of firms clustered in sectors indirectly linked to natural innovation fell in all the groups of firms considered. resources, on the other hand, the findings suggest that even However, innovative firms clustered in sectors linked in the darkest periods of innovation capacity destruction to natural resources were the ones that reduced their in the late 1980s and early 1990s, the agricultural sector investment in this area least as a percentage of sales. In sustained a group of chemical and machinery suppliers particular, innovative firms clustered in sectors directly that formed a locus of clustered innovative firms in the linked to natural resources were alone in practically vanguard of innovation activity as creators of knowledge maintaining their investment in capital goods for within the Argentine manufacturing sector. innovation at the levels of the start of the period, even It is also interesting to supplement the above data though their sales increased substantially. with others that reflect developments in firms’ innovation In summary, taken together for the whole period behaviour during the period analysed from 1992 to 2004. analysed, the two groups of innovative firms clustered Table 8 shows the findings for the indicators analysed. in sectors linked to activities based on the exploitation Over more than a decade, innovative firms clustered of natural resources constituted, by virtue of their in sectors both directly and indirectly linked to natural relative efforts in innovation activities, an important resources were distinguished by the growing intensity locus of knowledge creation and usage capacity in of r&d investment. In particular, firms in the former Argentine manufacturing industry. This was a substantial sectors increased the intensity of this spending almost and potentially important contribution, particularly three times as fast as the manufacturing sector average, at a time of great instability and upheaval like the i.e., by considerably more than non-innovative firms, one considered. which, along with this group of firms, were the ones that invested least in r&d at the start of the period. As 2. Clustered innovative firms in Brazil regards innovative firms clustered in sectors indirectly linked to natural resource-based activities, these increased Among clustered innovative firms in Brazil, whether the intensity of r&d spending more quickly than other or not linked to natural resource-based activities, local groups with a comparable intensity of r&d expenditure at business groups are more heavily represented and

TABLE 8 Argentina: changes in innovation behaviour in the manufacturing sector, 1992-2004

r&d intensity Capital investment

Types of firms Average values per firm (percentages)

1. Innovative firms clustered in sectors: 17.12 -4.29 - directly linked to natural resources 31.84 -0.38 - indirectly linked to natural resources 16.25 -8.96 2. All other innovative firms 8.69 -7.40 3. Firms in “high-technology” industries 10.83 -10.59 4. Subsidiaries of multinationals 11.02 -14.95 5. Non-innovative firms 23.41 -7.50 6. All firms 11.57 -7.34

Source: prepared by the authors, on the basis of the Survey on Technological Behaviour of Argentine Industrial Firms (1992-1996) and the National Survey of Firms on Innovation, Research and Development and Information and Communication Technologies (2002-2004). r&d: research and development.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 165 independent local firms less so than in the manufacturing as large an r&d effort as clustered firms not linked to sector as a whole. However, the composition of the group natural resource-based activities (1.17% and 0.52%, of innovative firms differs substantially depending on respectively). In addition, clustered innovative firms whether or not they are linked to natural resource-based linked to natural resources as suppliers rank alongside activities. Among those so linked, 44% are subsidiaries high-technology firms as those making the greatestr&d of foreign companies, a relatively strong presence, while efforts relative to sales in the year analysed. among those that are not, foreign firms are not heavily Observing the evolution of firms’ innovation activity represented (5%). In addition, local business groups indicators brings some other interesting findings to and independent local firms are more to the fore among light. Between 2000 and 2005, in a context where r&d natural resource-linked firms (45% and 40%, respectively) investment intensity generally declined in the Brazilian than among firms not linked to natural resource-based manufacturing sector, clustered innovative firms were the activities (23% and 33%, respectively), as can be seen only ones to increase their r&d effort year by year (see in table 9. In terms of size, clustered innovative firms are table 12). Clustered innovative firms not linked to natural larger on average than manufacturing firms generally, resource-based activities made most progress, increasing isolated innovative firms and non-innovators, and 87% r&d investment intensity by 16.15%. Other innovation of innovative firms clustered in natural resource-linked efforts, such as the acquisition of knowledge incorporated industries and 100% of clustered firms not linked to into capital goods, were generally negative during the natural resources have over 200 employees. Firms of period analysed. However, clustered innovative firms this size represent 56% of all firms in the manufacturing linked to natural resources showed the smallest decline sector (see table 10). in capital investment of any type of firm (4.64% a year). As regards innovation behaviour, the figures In summary, according to the indicators analysed, presented in table 11 show that clustered innovative clustered innovative firms in Brazil (whether or not linked firms as a group made greaterr&d efforts relative to to natural resource-based activities through input-output sales (0.91%) than none-innovative firms (0.36%) and the relationships) form an important locus of innovation in subsidiaries of multinationals (0.79%) in 2001, but made the Brazilian manufacturing sector. It is interesting to note less of an effort than other innovative firms (1.30%) and that these firms stand out throughout the period studied high-technology firms (2.06%). It should be noted that as the ones making the greatest innovation effort (such innovative firms clustered in sectors indirectly linked to as r&d), of which there is little in the manufacturing natural resource-based activities made more than twice sector of the region’s countries.

TABLE 9 Brazil: innovative firms by ownership type (Numbers and percentages)

Innovative firms Clustered innovative Non-innovative clustered in natural Isolated firms not linked to firms in Total resource-linked innovative firms natural resources all industries industries

Number of firms and proportions of column totals (percentages) Local business groups 7 9 29 486 522 (23) (45) (25) (13) (13) Independent local firms 10 8 43 2 677 2 730 (33) (40) (37) (72) (71) Subsidiaries of multinationals 13 3 45 559 617 (44) (5) (38) (15) (16) Total 30 20 117 3 722 3 889 (100) (100) (100) (100) (100)

Source: prepared by the authors, on the basis of the Survey on Technological Innovation (pintec) (2000 and 2005).

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 166 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 10 Brazil: firms by size (Numbers and percentages)

Innovative firms Clustered innovative clustered in natural Isolated Non-innovative firms firms not linked to Total resource-linked innovative firms in all industries natural resources industries smes 4 0 17 1 680 1 701 (13) (0) (15) (45) (44) Large firms 26 20 100 2 042 2 168 (87) (100) (85) (55) (56) Total 30 (100) 20 (100) 117 (100) 3 722 (100) 3 889 (100)

Source: prepared by the authors, on the basis of the Survey on Technological Innovation (pintec) (2000 and 2005). Note: small and medium-sized enterprises (smes) are those with 200 employees or less and large enterprises those with at least 200 employees.

TABLE 11 Brazil: innovation behaviour in the manufacturing sector, 2000

r&d intensity Capital investment

Types of firms Average values per firm (percentages)

1. Clustered innovative firms: 0.91 2.09 - indirectly linked to natural resources 1.17 1.74 - not linked to natural resource-based industries 0.52 2.61 2. All other innovative firms 1.30 3.56 3. Firms in “high-technology” industries 2.06 2.01 4. Subsidiaries of multinationals 0.79 6.03 5. Non-innovative firms 0.36 4.72 6. All firms 0.97 3.89

Source: prepared by the authors, on the basis of the Survey on Technological Innovation (pintec) (2000 and 2005). r&d: research and development.

TABLE 12 Brazil: changes in innovation behaviour in the manufacturing sector, 2000-2005

r&d intensity Capital investment

Type of firm Average values per firm (percentages)

1. Clustered innovative firms: 5.72 -12.99 - indirectly linked to natural resources 1.50 -4.64 - not linked to natural resource-based industries 16.15 -26.95 2. All other innovative firms -14.84 -10.80 3. Firms in “high-technology” industries -1.22 -8.66 4. Subsidiaries of foreign multinationals -3.89 -7.34 5. Non-innovative firms -2.14 -27.32 6. All firms -8.23 -12.72

Source: prepared by the authors, on the basis of the Survey on Technological Innovation (pintec) (2000 and 2005). r&d: research and development.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 167

VI Discussion and final reflections

This paper has studied the evolution of innovation for firms to exploit the advantages of technological capacity in the Argentine and Brazilian manufacturing change and add value, in comparison with other sectors during a period characterized by great economic activities such as industrial manufacturing (Prebisch, instability and changes in economic policies in both 1950; Singer, 1950; Nurske, 1958). Similarly, the countries. Many studies have documented the destruction innovation literature has identified manufacturing of firms and technological capacities that prevailed at industries closely linked to natural resources as this juncture in the region. What has been investigated having few technology opportunities; indeed, they here, though, is the possible resurgence of innovation are classified as “low-technology” industries in capacities and their sectoral distribution. The analysis industrial classifications oecd( , 1997). Consequently, has used intertemporal data from innovation surveys in the typical policy recommendation for countries Argentina and Brazil covering the period from 1992 to with a marked specialization in natural resources 2004 in the former and from 2001 to 2005 in the latter. has been that they should foster the development of The findings are not definitive because they are based on other sectors associated with greater opportunities analysis of such evidence as exists, which is imperfect to add value, such as those activities grouped into and fragmented. Nonetheless, they provide grounds for the so-called “high-technology” segments. Recently, some reflections and suggestions that are potentially though, some authors have begun to stress that important for future research. Two in particular are sectors based on the exploitation of natural resources worth mentioning and discussing. are intensifying their use of knowledge and thus First, a substantial number of clustered firms are opening up opportunities for greater learning and found in a small group of sectors, which were identified innovation, and thence for the development of as being at the vanguard of innovation and which greatly dynamic, innovative sectors associated with them increased their innovation efforts during the period (Marín, Navas-Alemán and Pérez, in press; Pérez, under study. The evidence analysed does not suggest 1999, 2001 and 2010; Kaplinsky and Fitter, 2004). that these firms are world leaders, one indication of this This is largely because of fundamental changes in being their low levels of r&d expenditure, which is well historical conditions, together with demand shifts below global standards; in relative terms, however, they and the spread of new technological paradigms. The stand out from other firms in each country, and it is thus demand for foodstuffs, raw materials and energy interesting to analyse their characteristics. has intensified in recent years due to the growing Second, it should be stressed that the activities globalization of markets (and the incorporation of many of these innovative firms are clustered in are not China into the world market), with a concomitant the ones which would typically be expected. Firms increase in the demand for variety and quality (such clustered in sectors that are users of natural resources as the increase in demand for gourmet foods and performed particularly well in Argentina, while this was safer or more environmentally friendly products). true of firms clustered in traditional sectors in Brazil This is opening up opportunities that did not and of firms linked to natural resources as suppliers formerly exist to invest in knowledge and generate and clustered in sectors protected by special regimes variety in conjunction with natural resources. in both countries. It is worth reflecting on the first two Meanwhile, the spread of new technologies such as types: (i) firms clustered in sectors linked to natural biotechnology and nanotechnology is multiplying resources, particularly as users, and (ii) firms clustered the potential for innovation and differentiation, just in traditional sectors. as information and communication technologies (i) Connection to natural resources: industries connected (icts) are favouring and facilitating the incorporation to natural resources are usually regarded as having of far-flung production zones into the global few technology opportunities. Since the 1950s, the market (Von Tunzelmann and Acha, 2005; Marín, development literature has argued that activities Navas-Alemán and Pérez, in press). The findings based on the exploitation and processing of natural presented here may be beginning to capture resources have characteristics that do not make it easy these phenomena.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 168 CEPAL REVIEW 114 • DECEMBER 2014

(ii) Traditional sectors: in the case of Brazil, areas of other associated activities that are usually less intensive competitive advantage creation can also be seen in r&d (Hauknes and Knell, 2009; Robertson and Patel, in sectors that are not closely linked to natural 2007). This interaction needs to be considered in the resources but are traditional, such as the manufacture region’s development strategies, avoiding excessive bias of footwear and plastic products. This may also be towards incentives for high-technology sectors that fail due to the penetration of new technologies in what to take account of their links to and interactions with are usually considered low- and medium-technology other production sectors. sectors (Von Tunzelmann and Acha, 2005) and to Lastly, some questions, limitations and suggestions other emerging phenomena that the literature has for future research may be noted: begun to identify. Recent studies have shown, in The indicators used have some limitations. First, fact, that traditional or low- to medium-technology those measuring innovation focus almost exclusively sectors are not necessarily clusters of non-innovative on the manufacturing sector, preventing investigation of firms (Kirner, Kinkel and Jaeger, 2009; Hirsch- innovation patterns in other activities that are important Kreinsen, 2008). This is because products typically to the region, such as primary activities and the service manufactured with old or mature technologies are sector. Again, the innovation outcome indicators available beginning to be produced with technologies that present limitations when it comes to accurately capturing are radically new or characteristic of some other the innovation dynamic in these countries. Patents are a industry (Kirner, Kinkel and Jaeger, 2009). The widely used and accepted indicator in more developed application of biotechnology to food processing is countries, but they have their limits, since the bulk of a clear example of how a sector regarded as being the innovation process in the region’s firms consists of non-technology-intensive has begun to adopt what innovations that are incremental and adaptive or have a are identified as high-technology solutions and low degree of novelty, which means that the patenting become increasingly dynamic. It is argued that this system cannot be used to protect them. Other innovation horizontal spread of technology between sectors outcome indicators available, such as the one used in this will tend to blur the identification of industries study, are based on subjective answers given by the firms with products and technologies and reduce the surveyed as to whether they innovated or not in a given usefulness of certain sectoral classifications that period and what degree of novelty was involved. The are very widespread and heavily used. literature recognizes that this may lead to overstatement Taken all together, these findings suggest that existing of firms’ innovativeness. Consequently, the limitations sectoral classifications may not be entirely relevant, or of the indicators used and of the data available require are not capturing recent phenomena such as intensified caution when it comes to interpreting the findings and innovation in what are traditionally considered low- making hard and fast generalizations. technology sectors, such as traditional ones and those Second, this paper speaks of a resurgence, but the linked directly to natural resources. They also suggest indicators of innovation efforts such as r&d are extremely that industrial development policies that involve picking low in some unexpected cases. It must be asked, then, winners and are centred on identifying and strengthening whether they are low because the indicators available isolated sectors may not be the most appropriate in the were unable to identify genuine innovators, or because current context. On the contrary, in industrial policy the most significant innovators are in natural resource and terms it seems more promising to consider the interaction traditional sectors where innovation is done differently and interdependence between sectors (Hansen and (incremental and process innovation and engineering Winther, 2010). It is known, for example, that firms in efforts are most important). More needs to be known what are usually deemed low- and medium-technology in future about the characteristics of innovation in industries are not only the most substantial in terms of each sector, with a view to identifying indicators that output, employment and investment, but are the largest serve to capture differences in the intensity of efforts consumers of high-technology innovations (Robertson, and outcomes irrespective of sectoral characteristics. Pol and Carroll, 2003). Consequently, the growth of high- This will certainly enable more light to be shed on the technology activities largely depends on the growth of questions raised in this study.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS CEPAL REVIEW 114 • DECEMBER 2014 169

ANNEX Identifying natural resource-related sectors

Two types of industries related to the production of The following procedure was applied for each natural resources are identified: consumers and suppliers. sector j: The classification was carried out using indices that measure the degree of connection to activities based on sum IOMm = natural resources,n = j 12 the exploitation of natural resources. These indices Index natural resourcesConsj = _ i sum IOMm,n = j use information on the value of transactions between _ i industries drawn from each country’s input-output matrix (iom). Because the activities in the iom do not unequivocally An analogous procedure was applied to create match the three-digit International Standard Industrial the second index for natural resource sector supplier Classification of All Economic Activities (isic), each industries. The greater the index value, the greater their sector’s transactions were weighted by the sum of the indirect connection to natural resources: value of production in the three-digit sectors corresponding to a given activity in the iom. T The first index is the proportion of purchases from sum IOMm = natural resources,n = j Index natural resources Sup = _ i natural resource sectors made by each manufacturing j T sum IOMm,n = j industry as a share of the sector’s total purchases. The _ i larger the proportion, the greater the direct connection to natural resources. The sectors identified as directly related to natural resources are those whose index values as consumers of these resources display above-average values, while 12 On the basis of the input-output matrix (iom), natural resource sectors sectors classified as indirectly related to natural resources are: growing of cereals, oilseeds and fodder; growing of vegetables, flowers and ornamental plants; growing of fruit and nuts; growing of are those whose index values as natural resource sector industrial crops; seed production; stockbreeding and production of suppliers are above average. Sectors with above-average milk, wool and hides; farm production; agricultural services; hunting; values in both groups were placed in the one for which forestry and timber extraction; fishing; oil and gas extraction and coal and uranium mining; mining of metal ores; mining of other minerals. they recorded the highest value.

Bibliography Amsden, A.H. (1989), Asia’s Next Giant: South Korea and Late Dahlman, C. and F. Fonseca (1987), “From technological dependence Industrialization, New York, Oxford University Press. to technological development: the case of Usiminas steelplant in Baer, W. (2001), The Brazilian Economy: Growth and Development, Brazil”, Technology Generation in Latin American Manufacturing Greenwood Publishing Group. Industries, J. Katz (ed.), London, Macmillan. (1972), “Import substitution and industrialization in Latin Dahlman, C. and C. Frischtak (1990), “National systems supporting America: experiences and interpretations”, Latin American technical advance in industry: the Brazilian experience”, Research Review, vol. 7, No. 1, Latin American Studies National Innovation Systems. A Comparative Analysis, R. Association. Nelson (ed.), Oxford, Oxford University Press. (1970), Siderurgia e desenvolvimento brasileiro, Rio de Erber, F., E. Guimarães and J. Tavares Araújo Jr. (1974), Absorção Janeiro, Zahar Editores. e criação de tecnologia na indústria de bens de capital, Rio Barros de Castro, A. (2003), “Brazil’s second catch-up: characteristics de Janeiro, Studies and Projects Financing Entity. and constraints”, cepal Review, No. 80 (LC/G.2204-P), Ferraz, J.C., D. Kupfer and L. Haguenauer (1996), “The competitive Santiago, Chile, August. challenge for Brazilian industry”, cepal Review, No. 58 Chudnovsky, D., A. López and G. Pupato (2006), “Innovation and (LC/G.1916-P), Santiago, Chile, April. productivity in developing countries: a study of Argentine Ferraz, J.C., D. Kupfer and F. Serrano (1999), “Macro/micro manufacturing firms’ behavior (1992-2001)”, Research Policy, interactions: economic and institutional uncertainties and vol. 35, No. 2, Amsterdam, Elsevier. structural change in the Brazilian industry”, Oxford Development Cimoli, M. and J. Katz (2003), “Structural reforms, technological gaps Studies, vol. 27, No. 3, Taylor & Francis. and economic development: a Latin American perspective”, Ferraz, J.C., D. Kupfer and M. Iootty (2004), “Industrial Industrial and Corporate Change, vol. 12, No. 2, Oxford, competitiveness in Brazil ten years after economic liberalization”, Oxford University Press. cepal Review, No. 82 (LC/G.2220-P), Santiago, Chile, April. Dahlman, C. (1984), “Foreign technology and indigenous Hansen, T. and L. Winther (2010), “Innovation, regional development technological capability in Brazil”, Technological Capability and relations between high- and low-tech industries”, document in the Third World, M. Fransman and K. King, London, presented at the 8th European Urban and Regional Studies Macmillan. Conference, Vienna.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS 170 CEPAL REVIEW 114 • DECEMBER 2014

Harabi, N. (1995), “Appropriability of technical innovations: López, A. (1996), “Las ideas evolucionistas en economía: Una visión an empirical analysis”, Research Policy, vol. 24, No. 6, de conjunto”, Revista Buenos Aires. Pensamiento Económico, Amsterdam, Elsevier. No. 1, Buenos Aires. Hauknes, J. and M. Knell (2009), “Embodied knowledge and sectoral Marín, A. and M. Bell (2012), “Argentina”, Innovative Firms in linkages: an input-output approach to the interaction of high- Emerging Market Countries, E. Amann and J. Cantwell (eds.), and low-tech industries”, Research Policy, vol. 38, No. 3, Oxford, Oxford University Press. Amsterdam, Elsevier. Marín A., L. Navas-Alemán and C. Pérez (in press), “Natural Hirsch-Kreinsen, H. (2008), “‘Low-tech’ innovations”, Industry and Resource Industries as a Platform for the Development of Innovation, vol. 15, No. 1, Taylor & Francis. Knowledge Instensive Industries”. Hobday, M. (1995), Innovation in East Asia: The Challenge to Japan, Mathews, J. and D.-S. Cho (2000), Tiger Technology: The Creation of Aldershot, Edward Elgar. a Semiconductor Industry in East Asia, Cambridge, Cambridge Kaplinsky, R. and R. Fitter (2004), “Technology and globalisation: University Press. who gains when commodities are decommodified?”, Nurske, R. (1958), “Trade fluctuations and buffer policies of low- International Journal of Technology and Globalization, vol. 1, income countries”, Kyklos, vol. 11, No. 2, Wiley. No. 1, Inderscience Publishers. oecd (Organization for Economic Cooperation and Development) Katz, J. (2007), “Cycles of creation and destruction of production (2005), Oslo Manual: Guidelines for Collecting and Interpreting capacity and the development of new export and technological Innovation, Paris. capabilities in Latin America”, Can Latin American Firms (1997), “Revision of the high-technology sector and Compete?, T. Grosse and L. Mesquita (eds.), Oxford product classification”, oecd Science, Technology and Industry Scholarship Online. Working Papers, No. 1997/02, Paris. (2001), “Structural reforms and technological behaviour. Patel, P. (2000), “Technological indicators of performance”, From The sources and nature of technological change in Latin Knowledge Management to Strategic Competence: Measuring America in the 1990s”, Research Policy, vol. 30, No. 1, Technological, Market and Organisational Innovation, J. Tidd Amsterdam, Elsevier. (ed.), London, Imperial College Press. (1987), “Domestic technology generation in ldcs: a Pérez, C. (2010), “Technological dynamism and social inclusion review of research findings”, Technology Generation in Latin in Latin America: a resource-based production development American Manufacturing Industries, London, Macmillan. strategy”, cepal Review, No. 100 (LC/G.2442-P), Santiago, (1972), “Importación de tecnología, aprendizaje local e Chile, April. industrialización dependiente”, Documento de trabajo, Buenos (2001), “Technological change and opportunities for Aires, Instituto Torcuato di Tella. development as a moving target”, cepal Review, No. 75 Katz, J. and N.A. Bercovich (1993), “National systems of innovation (LC/G.2150-P), Santiago, Chile, December. supporting technical advance in industry: the case of Argentina”, (1999), “Cambio de patrón tecnológico y oportunidades para National Innovation Systems: A Comparative Analysis, el desarrollo sustentable”, Colección Ideas para el Diálogo, R.R. Nelson (ed.), New York, Oxford University Press. No. 3, Caracas, Biblioteca Nacional de Venezuela. Katz, J. and B. Kosacoff (2000), “Import-substituting industrialisation Prebisch, R. (1950), The Economic Development of Latin America in Argentina, 1940-1980: its achievements and shortcomings”, and Its Principal Problems (E/CN.12/89), New York, Industrialisation and the State in Latin-America: the Post United Nations. War Years, E. Cárdenas, J.A. Ocampo and R. Thorp (eds.), Robertson, P.L., E. Pol and P. Carroll (2003), “Receptive capacity Basingstoke. of established industries as a limiting factor in the economy’s (1989), “El proceso de industrialización en la Argentina: rate of innovation”, Industry and Innovation, vol. 10, No. 4, Evolución, retroceso y prospectiva”, Buenos Aires, Centro Taylor & Francis. Editor de América Latina. Robertson, P.L. and P.R. Patel (2007), “New wine in old bottles – Kim, L. (1997), Imitation to Innovation: The Dynamics of Korea’s technological diffusion in developed economies”, Research Technological Learning, Boston, Massachusetts, Harvard Policy, vol. 36, No. 5, Amsterdam, Elsevier. Business School Press. Rocha, F. and D. Kupfer (2002), “Structural changes and specialization Kirner, E., S. Kinkel and A. Jaeger (2009), “Innovation paths and the in Brazilian industry: the evolution of leading companies and innovation performance of low-technology firms - An empirical the M&A process”, The Developing Economies, vol. 40, No. 4, analysis of German industry”, Research Policy, vol. 38, No. 3, Tokyo, Institute of Developing Economies, December. Amsterdam, Elsevier. Scherer, F.M. (1983), “The propensity to patent”, International Kosacoff, B. (2000a), Corporate Strategies under Structural Journal of Industrial Organization, vol. 1, No. 1, Amsterdam, Adjustment in Argentina: Responses by Industrial Firms to a Elsevier. New Set of Uncertainties, Basingstoke, Macmillan. Singer, H. (1950), “The distribution of gains between investing and (2000b), El desempeño industrial argentino más allá de borrowing countries”, American Economic Review, vol. 40, la sustitución de importaciones, Buenos Aires, eclac office No. 2, Nashville, Tennessee, American Economic Association. in Buenos Aires. Teitel, S. (1981), “Towards an understanding of technical change in (1996), “Estrategias empresariales en la transformación semi-industrialized countries”, Research Policy, vol. 10, No. 2, industrial argentina”, Working Paper, No. 67 (LC/ BUE/L.150), Amsterdam, Elsevier. Santiago, Chile, Economic Commission for Latin America and Teitel, S. and F. Thoumi (1987), “De la sustitución de importaciones the Caribbean (eclac). a las exportaciones: La experiencia de las exportaciones Kulfas, M., F. Porta and A. Ramos (2002), “Inversión extranjera y manufactureras de la Argentina y el Brasil”, Desarrollo empresas transnacionales en la economía argentina”, Estudios Económico, vol. 27, No. 105, Buenos Aires, Institute of y Perspectivas series, No. 10 (LC/L.1766-P), Buenos Aires, Economic and Social Development. eclac office in Buenos Aires. United Nations publication, United Nations (1969), “The manufacture of machine tools in Sales No. S.02.II.G.80. Brazil”, Development of Metalworking Industries in Developing Levin, R.C. and others (1987), “Appropriating the returns from Countries, New York. industrial research and development”, Brookings Papers on Von Tunzelmann, N. and V. Acha (2005), “Innovation in ‘low-tech’ Economic Activity, vol. 18, No. 3, Washington, D.C., The industries”, The Oxford Handbook of Innovation, J. Fageberg, D. Brookings Institution. Mowery and R. Nelson (eds.), Oxford, Oxford University Press.

TECHNOLOGICAL CAPACITY-BUILDING IN UNSTABLE SETTINGS: MANUFACTURING FIRMS IN ARGENTINA AND BRAZIL • ANABEL MARÍN, LILIA STUBRIN AND MARÍA AMELIA GIBBONS Index of political instability in Brazil, 1889-2009

Jaime Jordan Costantini and Mauricio Vaz Lobo Bittencourt

ABSTRACT This article aims to develop an index of political instability (ins) in Brazil between 1889 and 2009, reflecting a wide-ranging set of multiple phenomena that represent conflicts between the different social groups. By presenting different definitions of what is understood by political instability in the economics literature and by using multiple historical events —coups d’état, civil conflicts, constitutional or unconstitutional overthrow and changes in the composition of 50% of the ministerial cabinet— different indicators are obtained which are then synthesized into a single index using the principal component technique,

to obtain an ins for Brazil between 1889 and 2009.

KEYWORDS Political history, political conditions, measurement, statistical methodology, statistical data, Brazil

JEL CLASSIFICATION N00, O54, P16

AUTHORS Jaime Jordan Costantini is a member of the research group of the International Economy and Economic Development Unit (neide) of the Federal University of Paraná (ufpr), Brazil. [email protected] Mauricio Vaz Lobo Bittencourt is a Professor for the Graduate Studies Program in Economic Development of the Federal University of Paraná (ufpr), Brazil. [email protected] 172 CEPAL REVIEW 114 • DECEMBER 2014

I Introduction

To gain a better understanding of the economic and Wen (1998), and Woo (2005), who apply economic development process in individual countries, the economics theory to study the effects of power conflicts in societies literature generally uses theoretical and empirical models that are unequal and ethnically polarized or divided. that do not have a clear definition of the important role of The third approach involves empirical research. political instability in the development process. Moreover, The key studies in this category include those of Barro no account is taken of the historical role of instability, (1991 and 1996), who used panel-data techniques to and the empirical studies rely on ceteris paribus-type study the effects of political instability on economic assumptions to analyse their results. growth in samples of countries. Several recent studies It is extremely important not only to find ways to use the generalized method of moments (gmm) to deal make studies of economic development more realistic, with endogeneity problems, including Aisen and Veiga but also to provide stronger support to the empirical (2011) who attempt to identify the channels through analyses undertaken, with theoretical, empirical and which political instability manifests itself in the data models that fill the shortcomings in question —in economic structure. particular the difficulty of incorporating institutional and Other empirical studies are of a regional type and historical variables into the analyses—. Such variables can analyse the topic in relation to a set of countries. The manifest themselves in the country’s political stability, or leading examples include Solimano (2003), who analysed instability, which merely reflect processes of institutional political instability in Andean countries, using a variety change over time —processes that simultaneously affect of variables (change of constitution, presidential crises and are affected by economic variables. and volatility of democracy) to explain the deficient Economists are very keen to understand the relations performance of their institutions. There are also national that exist between political instability and economic studies, such as Evia, Laserna and Skaperdas (2008) on the activity in individual countries. There has been much effects of social protests in the Bolivian economy, which theoretical and empirical research on the economic effects caused a drop in output of several percentage points, and of instability, and the main studies can be summarized Muñoz (2009), who found that political instability is one in terms of three approaches. of the main causes of the declining trend of economic The first approach studies the effects of political activity in the Bolivarian Republic of Venezuela. instability on the economy by explaining fiscal-policy The aim of the present study is to fill some of cycles. Studies of this type include those of Rogoff the gaps mentioned above, by calculating an index of and Sibert (1988); Alesina and others (1992); Alesina political instability (ins) in Brazil between 1889 and and Tabellini (1990); Edwards and Tabellini (1991); 2009. An index of this type will make it possible to Cukierman, Edwards and Tabellini (1992), and Bohn analyse the main trends of political instability in Brazil’s (2003). These highlight three main effects of political history, and to relate political phenomena to those that instability on the economy: (i) political instability are strictly in the economic domain. The index could generates suboptimal public policies since policymakers be used in a variety of future studies to gain a better are uncertain of their own permanency in the government understanding of the Brazilian development process, and they manipulate the macroeconomy as a tool in because it embraces the political, social, historical and the struggle for power; (ii) entrepreneurial investment institutional dimensions as well as the economic one. decisions are hampered by a climate of political instability This article is organized into six sections including in which governments implement suboptimal policies, the introduction. Section II discusses the different and (iii) this instability causes interruptions in production, definitions of political instability and appropriate methods which reduces total factor productivity (tfp). for measuring it. Section III describes the historical The second approach investigates the relations that context of political instability in Brazil, and section exist between political instability and economic growth, IV presents the methodology used. Section V includes by modelling power conflicts between the social classes the results and the time series of political instability in and the effects of such conflicts on capital accumulation. Brazil between 1889 and 2009, while the last section Research in this category includes Annett (2001); Devereux contains the final thoughts of the article.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 173

II What does political instability mean?

The specialist literature contains the two definitions of by a broad and diverse set of phenomena that affect political instability. The first of these sees instability as the that instability. propensity for the government to be replaced before the Episodes of political instability in Brazil have often end of its legal mandate, whether through constitutional been linked to social and political changes such as those or unconstitutional means. This definition is applied by that occurred in the so-called Vargas Era. In such cases, Cukierman, Edwards and Tabellini (1992). the elements of political instability are associated with The second definition, used by Annett (2001) and a varied set of multiple political and social phenomena, Aisen and Veiga (2011), among others, views political which makes the second definition of political instability instability as resulting from a set of multiple phenomena more appropriate. representing conflicts between different social groups. When political instability is related to multiple The phenomena in question could be civil wars, coups factors, statistical techniques need to be applied to develop d’état, violent protests or new constitutions, or other a single variable representing the political instability of events; and they are generally measured as categorical the set of factors in question. The appropriate statistical variables. This definition of political instability offers a technique for that purpose is principal component better representation of the country’s political evolution, analysis, which has been applied by Annett (2001) and because it reflects a complex process determined Aisen and Veiga (2011).

III Historical context of political instability in Brazil between 1889 and 2009

At the close of the nineteenth century, Portuguese based and “capitanias” (colonial land management colonialism had bequeathed four organizational legacies districts) or land grants to the colonists.2 that were to prove very persistent and would characterize (iii) Colonialism and mercantilism as the defining the profile of institutions in Brazil: characteristics of Brazil’s international engagement. (i) The delegation of significant functions to the local- The colonies had to contribute to the economy of authority level (Fausto, 1996), thereby perpetuating the metropolis in certain areas, with the exclusion of the pragmatism of the colonial power in the Brazilian other colonial powers.3 Surpluses for export came empire. Article 66 of the Constitution of 24 February from forced labour obtained from the indigenous 1822 expressly authorized the states to borrow, population and, later, from negro slavery. The organize their own military forces, and impose cycles of mercantilism were associated with certain taxes by decree on their merchandise exports. regions of Brazil and specific export products, such (ii) A high concentration of land ownership, structured in large extensive areas,1 an institution that originated in the later colonialism of the hereditary “sesmaria”- 2 The creation of the sesmarias gave rise to the latifundio or estate system, which concentrated political power and a governing class in the economy in a way that would have lasting effects in Brazil. 3 In the case of Brazil, colonial exclusiveness was less strict because 1 According to Fausto (1996, pp. 24-26), although the discovery of of the impossibility of imposing it and the fact that the trade centres Brazil by Pedro Álvarez Cabral occurred in 1500, colonization as such were outside Portugal. In the sixteenth century, the Netherlands began in 1549; and it was consolidated in 1763 with the foundation participated in the colonial benefits, and, in the seventeenth century, of the first capital in the city of Salvador, Bahia. England also entered the colonial scheme.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 174 CEPAL REVIEW 114 • DECEMBER 2014

as Pernambuco (in the sixteenth century); exports international bankers that participated in the plan to of sugar from the North-East (in the sixteenth defend the coffee price demanded greater rigour in the century, supported by the Netherlands); or the State management of public resources to guarantee credit of Minas Gerais with exports of gold and diamonds solvency. Moreover, the states that did not participate (in the seventeenth and eighteenth centuries, as a in the oligarchic pact had different interests than key development for Portugal).4 Other commodities those of the coffee growers of the state of São Paulo. (tobacco and rubber) became alternative crops; Consequently, during that period, unorthodox economic and coffee production began in the mid-nineteenth policies alternated with more liberal ones, reflecting the century, initially in Rio de Janeiro and later in São concern for fiscal balance, inflation, and the viability of Paulo, which would trigger major transformations the balance of payments. in Brazil. Management of the oligarchic pact was rendered (iv) A tradition of forced labour from Indians and more complex by the emergence of both economic Afro-descendants, which came to an end in 1880, and political problems. The most significant economic so the Brazilian power elites had to encourage problem was the crisis of 1929, which had effects on immigration by foreigners for coffee growing. two fronts. As from 1926, Brazil found it more difficult According to Maddison (2008), between 1880 and to obtain the external credits needed to maintain the 1913, a total of 2.7 million foreigners arrived in coffee price, owing to the restrictive policies that had Brazil (half of them Italians, 700,000 Portuguese and been implemented in the United States to control the 400,000 Spanish, in addition to smaller numbers of speculative bubble on Wall Street. In 1932, coffee prices other nationalities). For a country that had 17 million were 30% lower than in 1929, and coffee represented inhabitants at the turn of the twentieth century, those about 70% of Brazil’s exports. The political problem migration figures are highly significant. was the breakup of the oligarchic pact which enabled the leader of the State of Rio Grande do Sul, Getúlio 1. The First Republic and the Oligarchic Pact Vargas, to seize power with a military coup. between 1889 and 1929 2. Economic changes between 1930 and 1964 The concentration of land ownership, in conjunction with the presence of a skilled immigrant labour force In 1930, under Vargas’ leadership, a reform process and opportunities for coffee production in São Paulo, was launched in the political organization of Brazil, fuelled the desire to establish a political power structure which would last 15 years. The fundamental element that favoured the interests of the power elites. Another of the reforms in question involved the centralization factor was the new Liberal Constitution of 1891, which of government power, which brought the states’ high gave the states autonomy from federal power, thereby level of autonomy to an end. The reforms gave rise to making it possible to introduce measures to enhance conflicts that led to the outbreak of the Constitutionalist coffee growing, which was fundamental for the State Revolution (or Paulista war) of 1932. The structure of of São Paulo. the state was reformed by expanding its sphere of action The oligarchic pact between the political elites in the economy, and numerous public institutions were was based on coffee and the spread of urbanization in created. The new laws gave broader recognition to São Paulo, Rio de Janeiro, Belo Horizonte, Bahia and workers’ rights, and the minimum wage was established; Porto Alegre. Politically, the elites of São Paulo and but the right to strike was heavily restricted. The role of Minas Gerais held national power alternately, under the industry in the economy was highlighted with the creation so-called “café com leite” politics based on the interests of the National Iron and Steel Corporation (Compañía of coffee-growers and dairy farmers, respectively. Siderúrgica Nacional) in Volta Redonda, and tools were The preponderance of coffee-grower interests created to promote industrialization through the Brazilian in defining the economic policy of the time gave an Development Bank (bndes). The authoritarian rule of unorthodox economic orientation to public affairs, Getúlio Vargas between 1937 and 1945, consolidated owing to the expansionary effects on public expenditure changes that would have lasting effects in Brazil. During of policies to sustain the coffee price. Nonetheless, the the World War II, difficulties in obtaining goods and raw materials from abroad provided favourable conditions for an import substitution policy, which contributed to 4 Portugal’s trade deficit with England was paid with gold that came the country’s modernization. from Minas Gerais.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 175

With the end of the Vargas government, the 4. Democratic government since 1985 Third Republic was inaugurated along with the period of populist democracies that would last from 1945 The transition to democracy was a complex and lengthy to 1964. During that period, the Vargas legacy was process, owing to at least three challenges among the consolidated in the form of pro-industrialization policies issues included on the political agenda. The economic with robust state support. Between 1945 and 1961, the problem was one of these and seemed to be the most economy grew rapidly thanks to a positive international urgent. The second was the need perceived by large context (Abreu, 2000). Nonetheless, economic trends segments of society for a new constitution to replace the changed5 during the government of João Goulart (1962- existing one, whose clearly authoritarian characteristics 1964); and in that period, various political problems were the legacy of military rule. The third challenge arose, which culminated in the military coup of was to address the heavy social debt in terms of income March 1964. distribution, which had resulted from the economic policies applied in the two decades of military rule —a 3. Military rule between 1964 and 1985 debt that needed to be resolved to smooth the harsh contours of Brazil’s social reality—. It should be noted The historical context of the different governments that that the working class played a major role in promoting succeeded each other in power between 1964 and 1985 the democratic process, thanks to its high degree of can be divided into three stages: (i) the first, between organization and power, which could not be ignored by 1964 and 1967, was characterized by the attempt to the Brazilian political class. achieve basic equilibrium in the economy (mainly in Although the Brazilian political system in 1985 terms of inflation); (ii) the economic-miracle period from involved the main leaderships and political powers in the 1967 to 1979, and (iii) a period of stagnation between country and had democratic institutions, it was unable to 1979 and 1985. address these three challenges rapidly and simultaneously. The Government’s Program of Economic Action Describing the paths of the solutions implemented for (paeg), between 1964 and 1967, was an inflation control each challenge goes beyond the scope of this article, plan to correct the balance of payments deficits. During but the solutions finally adopted by Brazilian society those years, inflation was brought under control gradually, are briefly summarized below. as gdp grew moderately. In the economic-miracle years In 1988, a new democratic Constitution was from 1967 to 1979, per capita gdp grew by 6.2% per approved, which reclaimed the legacy of the Vargas Era. year, despite external problems caused by the rise in oil A major role was given to the State both in the economy prices, and an aggravation of the situation after 1982 and in society; and equity criteria were established in when high international interest rates made it difficult relation to gender, income and regional differences. to finance the balance of payments. During this period, The process of solving the country’s economic deep domestic imbalances were compounded by high problems was long and tortuous; and there were six failed domestic inflation and a perception of fragility in the economic stabilization plans before Brazil found the alliances that provided the political underpinning for way to overcome the problem of inflation in 1994 with military rule. As a political response, a democratic the Real Plan. This solution was ultimately consolidated transition process was launched, starting with the in President Fernando Henrique Cardoso’s second term government of President Geisel in 1979. This would of office. Nonetheless, even with democracy, output culminate in the handover of power to a civilian growth remained unsatisfactory, because, between government under an indirect electoral system introduced 1980 and 2008, per capita gdp grew by just 0.77% per by the last military president, João Baptista Figueiredo, year, which suggests the presence of obstacles to rapid in 1985. growth in Brazil. In the social domain, significant progress was made, thanks to government policies that reduced extreme poverty; but progress on income distribution was limited, and public policies still have a long way to 5 Gross domestic product (gdp) grew by 8.6% in 1961; but in 1963 go. Major investments are also required in the areas of growth was just 0.63%, and industrial output shrank by 0.2%. Inflation, health, education, infrastructure and security; and the measured by the General Price Index-Domestic Supply (igp-di)/ Getúlio Vargas Foundation (fgv), was 30.5% in 1960 and 47.8% in historical need for reforms in both the political and tax 1961, and it reached a level of 92.1% in 1964. systems still persists.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 176 CEPAL REVIEW 114 • DECEMBER 2014

IV Methodology

The methodology applied includes definition of the mandate, the taking of office of the president-elect, variables, information sources, the resulting databases, changes in the cabinet, the approval of new constitutions and the statistical techniques used to synthesize political and reform of existing ones, civil wars, and the number instability in Brazil in a single index. of strikes. The annex includes a detailed list of the information sources used. 1. Definitionof the variables 2. Basic concepts of the principal Political instability is measured through several indicators component technique and is considered to exist in a given year if any of the following conditions are fulfilled: Using data on the variables defined above (nc, mg, mc, (i) Interruption of the President’s constitutional gc and greve), the multivariate principal component mandate (nc). In any given year, the constitutional technique is applied to obtain a synthetic vector of mandate is not fulfilled, owing to the resignation, political instability in Brazil in the period 1889-2009. constitutional or otherwise, of the President of the According to Gurmu, Rilstone and Stern (1999), Republic. In this situation, the president-elect does the principal component technique is a multivariate not take office owing to an unconstitutional act or statistical approach that reduces a large number of some other event. It occurs when the president- variables to fewer dimensions. It is particularly useful elect dies before taking office, or owing to a force when the variables considered are correlated. The principal majeure event that prevents the President from component method creates uncorrelated indices. From taking office, or when the President is unable to a mathematical point of view, with an initial set of n fulfil his or her mandate because of a coup. correlated variables, the principal component technique (ii) A change in 50% the ministerial cabinet (mg). This creates uncorrelated components, each of which consists variable represents the number of times in the year of a weighted linear combination of the initial variables. that the head of government changes half of the For example, for a set of variables X1, … …, Xn, the ministerial cabinet. following components are defined: (iii) Change of constitution (mc), owing to the approval of a new constitution or constitutional reforms. CP = a X + a X + … + a X (1) (iv) Civil wars or organized acts of violence for political 1 11 1 12 2 1n n purposes (gc). Violence is deemed to be organized CP = a X + a X + … + a X (2) if more than 100 citizens participate in it. This is 2 21 1 22 2 2n n a variable measured dichotomously: gc = 1 in the CP = a X + a X + … + a X (3) event of a civil war, acts of violence, or irregular n n1 1 n2 2 nn n military forces; otherwise gc = 0. (v) Labour strikes (greve). This is measured as the where amn represents the weight of the principal component number of strikes, normalized on the interval [0.1]. m of the variable n. Each principal component (cp) is This article makes an innovation with this variable weighted by the auto-vector of the correlations matrix compared to the main known indices, which are if the original data are normalized. The variance of each generally based on the number of political strikes. cp is the eigenvalue corresponding to each eigenvector. The innovation reflects the fact that workers’ rights The components are ranked from the first component in Brazil were restricted for many years (during cp1 (1), which explains most of the original value, the Vargas Era as from 1937 and, in recent times, subject to the constraint that the sum of the squares between 1964 and 1973). After those periods, there of the weights must be equal to one, in other words, 2 2 2 were many strike movements in which it is hard to a 11 + a 12 … + a 1n = 1. The second principal component differentiate the political and economic factors. cp2 (2), explains a smaller proportion of the variation of To determine the value of the five indicators the original value and is not correlated with (1). Each cp mentioned, information is needed on the constitutional captures a smaller proportion of the principal. This makes

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 177

T it possible to reduce to a single index, using the eigenvalues minimising tr(X – p1a1 – p2a2) (X – a1p1 – p2a2), where a2 of the corresponding eigenvector as the weighting factor. is the eigenvector associated with the second eigenvalue. Technically, as the principal component describes This second component represents the factors that are a set of variables in terms of another smaller set of not represented by the first one. Following this logic, variables, the method identifies “n” linear combinations of each cp captures the variance of the previous one, which the “n” columns of the XTX matrix (XT is the transposed will be represented by its eigenvalue. It can be shown matrix), which are mutually orthogonal and have the that the contribution to reducing the variability of the T property that the first principal componentp 1 minimises cp is tr(X) (X) – δ1, where δ1 is the highest eigenvalue. T tr(X – p1a1) (X – a1p1), where tr denotes trace and a1 is The following section describes the application the eigenvector of XTX associated with the eigenvalue. and analysis of the technique to produce an indicator The second principal component p2 is obtained by of political instability for Brazil.

V Results

This section presents the descriptive statistics of the (nc) and changes in the Constitution (mc). This finding database used, the analyses made of the results of may indicate that changes to existing constitutions and the principal component and, lastly, an analysis of the the creation of new constitutions in Brazil were preceded behaviour of Brazilian political instability. by changes of government outside of a legal mandate. It could also mean that constitutional reforms are seen as 1. Descriptive statistics acts that create a new order, owing to political movements that use force, the main exception to this being the 1988 The results include preparation of the matrix of the values Constitution. There is a negative correlation between of the variables for all years considered (see annex A.6). episodes of organized violence (gc) and strikes (greve) The correlations matrix shown in table 1 serves as an As noted above, strikes tended to be recurrent when illustrative indicator to analyse the relations between workers rights were not fully upheld in previous periods. the indicators used. Table 2 reports the results of the application of the principal component method in the decomposition of the eigenvalues and eigenvectors. The eigenvectors are TABLE 1 Brazil: matrix of correlations between the presented in their orthonormal form, uncorrelated and variables related to political instability, normalized, with a dimension that is compatible with 1889-2009 the five variables used. mg nc mc gc greve It can be seen that cp1 explains a major share of the sample variance. In terms of the eigenvectors, the most mg a 1 nc b 0.0065 1 important variables are the creation of new constitutions mc c 0.1587 0.3115 1 and constitutional reforms, and the interruption of gc d -0.0038 -0.0804 0.1059 1 constitutional mandates. greve e 0.1608 0.0252 -0.0901 -0.2416 1 The first principal component was used to obtain Source: prepared by the authors. the political instability index (ins), giving rise to the a mg indicates changes of 50% of the composition of the ministerial following system: Cabinet. b nc refers to interruption of the President’s constitutional mandate. c mc reflects changes in the Constitution. ins = 0.2781*mg + 0.5982*nc + 0.7260*mc + d (4) gc refers to civil wars and organized acts of violence. 0.1712*gc – 0.1352*greve e greve refers to strikes. Note: the total number of observations is 120 and the significance level is 5%. From the result of the estimations, an ins is obtained for each year (see table A.6). The data show a significant positive correlation Figure 1 shows the trend of this index between between the interruption of the constitutional mandate 1889 and 2009.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 178 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 2 Brazil: correlation of the principal components of political instability, 1889-2009 Component Eigenvalue Difference Proportion Cumulative Component 1 1.35551 0.04836 0.27110 0.27110 Component 2 1.30714 0.27947 0.26140 0.53250 Component 3 1.02768 0.31784 0.20550 0.73800 Component 4 0.70984 0.11000 0.14200 0.88000 Component 5 0.59983 - 0.12000 1.00000

Principal components (eigenvectors)

Variable cp1 cp2 cp3 cp4 cp5 Not explained mg a 0.2781 0.3575 0.7532 -0.3190 0.3594 0 nc b 0.5982 0.1967 -0.4888 0.2140 0.5703 0 mc c 0.7260 -0.0246 0.0400 -0.0819 -0.7812 0 gc d 0.1712 -0.6031 0.4217 0.6371 0.1524 0 greve e -0.1353 0.6849 0.1200 0.6632 -0.2417 0

Source: prepared by the authors. a mg indicates changes of 50% in the composition of the ministerial Cabinet. b nc refers to interruption of the President’s constitutional mandate. c mc reflects changes in the Constitution. d gc refers to civil wars and organize acts of violence. e greve refers to strikes. Note: the total number of observations is 120; components = trace = 5.

FIGURE 1 Brazil: index of political instability (ins), 1889-2009

1.6 1 2 3 4 5

1.4

1.2

1.0

0.8

0.6

Index of political instability 0.4

0.2

0.0 1889 1909 1929 1949 1969 1989 2009

-0.2 Source: prepared by the authors (see annex A.6). Note: periods of high political instability are highlighted, numbered from one to five.

Figure 1 shows the ins for Brazil between 1889 2. Analysis of political instability and 2009. The higher the index value, the greater is political instability. The ins index was smoothed using Figure 1 indicates five periods of political instability in the three-year moving average to identify the major Brazil, which are analysed in the following paragraphs. trends in political instability in Brazil, and periods of 1889-1894. The First Republic was initially major instability are highlighted. accompanied by a period of major political instability. The analysis of figure 1 shows that Brazil had President Marechal Deodoro da Fonseca dissolved periods of great political instability but those years were Parliament, in violation the Constitution that had just not predominant in the country’s history. been promulgated in 1891. This measure was the result

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 179 of disagreements between Deodoro da Fonseca and the the Copacabana Fort revolution in 1922 and, subsequently, coffee-growing elite. In the end, the President resigned the Coluna Prestes revolt. A state of siege prevailed to avoid a civil war, and Vice President Floriano Peixoto, throughout the Bernardes presidency, with the consequent another military man, took office as the second President restrictions on political freedoms. of Brazil in 1892. Under the Constitution of 1891, The Vargas Era (1930-1945). The changes that elections should have been convened nine months later, occurred in Brazil during this period represented a but this did not happen. As a result, President Peixoto response to “café com leite” politics by the elites of faced several organized military uprisings, but managed certain states (particularly Rio Grande do Sul and to overcome all rebellions until March 1894, when his Paraíba), which were competing for power. The period term of office came to an end. known traditionally as the Vargas Era encompassed the At the same time as these conflicts were unfolding, following administrations: (i) the provisional government the Federalist Revolution occurred in the southern state of 1930-1934, which emerged from the civil and military of Rio Grande do Sul, which caused much bloodshed uprising of 1930 and gave rise to the new Constitution of and ended in 1895. Lastly, between 1895 and 1897 the 1934; (ii) the government of President Vargas between War of Canudos also caused a large number of fatalities. 1934 and 1937, of democratic type, resulting from The economic context in that period was characterized popular election; (iii) the Estado Novo (New State) by a significant downward trend in the price of rubber dictatorship from 1937 to 1945, during which a new (then the country’s main export product). constitution was approved in 1937, and which ended Following a period of political instability, the military in a coup d’état in 1945, and (iv) the constitutional gave up power and recognized the electoral victory of government of Getúlio Vargas, which began in 1951 Prudente de Morais, a coffee grower from São Paulo, and ended with his suicide in 1954. The Vargas Era was thereby launching the period of “café com leite” politics, based on a centralized state and strong executive, which in which the political leaders of São Paulo and Minas profoundly changed the “oligarchy pact” of the First Gerais alternated in power. Republic, and constituted a significant source of conflict. These phenomena together make it possible to President Vargas used the new Constitution of 1937 as a identify a period of high political instability at the start mechanism to institutionalize that change, while closing of the First Republic. The index prepared records a down Congress and decreeing that governors would be mean of 0.601, with a standard deviation of 0.271 and subject to the President’s authority. a maximum of 0.997. There were episodes of instability throughout that 1895-1930. These years are not highlighted as a period, including the civil and military uprising of 1930, period of extreme political instability in figure 1, despite the constitutional Paulista War of 1932, the installation quite intense instability related to certain economic of the Estado Novo dictatorship in 1937, the military factors. From the standpoint of those factors, the start of overthrow of Getúlio Vargas in 1945, and his subsequent the new century heralded a recovery in the international suicide in 1954. economy, which had a positive effect on Brazil. Castelar Political instability between 1945 and 1964. These and others (2001) show that the investment rate in Brazil years are known as the period of populist democracy, rose from 4.9% of gdp in 1901 to 17.8% in 1909. Much owing to the succession of governments that maintained of that investment was channelled into infrastructure and the legacy of the Vargas Era involving development manufacturing industry. Between 1901 and 1914, the supported by the State, in an international scenario that railway network expanded by 4%, and electric power was propitious for policies of that type. This enabled generation capacity grew by 13 times. During the World Brazil to achieve a per capita gdp growth rate of around War I, however, investment retreated and the terms of 4.5% per year, along with a substantial degree of industrial trade declined. Then, between 1918 and 1922, that diversification. At the end of the government of Juscelino situation was reversed, when Brazilian exports doubled Kubitschek, signs of economic imbalance appeared in two years. In 1923, global business cycles changed in the form of high inflation rates, and in 1961 Brazil again, and Brazil faced a balance of payments crisis and entered a process of political instability that intensified high inflation (31% per year), which led to orthodox in the ensuing years. economic policies being implemented throughout the 1961-1970. This was the longest-lasting period presidency of Artur Bernardes. of political instability in Brazil, in which the Brazilian At the same time, dissatisfaction among middle- Constitution was changed five times. Although there ranking military officers was reflected in events such as were phases of rapid economic growth, the profound

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 180 CEPAL REVIEW 114 • DECEMBER 2014 difficulties in dealing with the major economic imbalances ideal” and protected Afro-descendants and women from in Brazil were also clear to see. Political instability discrimination. Between 1985 and 1995, ministers of began with the resignation of Jânio Quadros in 1961, finance were very unlikely to still be in their post one and his replacement by Vice President João Goulart. The year after taking office. As a result, Brazil had 11 finance government of President Goulart suffered from internal ministers6 in 10 years. The political instability in 1992 instability and continuous changes in its ministerial was caused by the constitutional overthrow of President composition. At the same time, there was growing Fernando Collor following an impeachment process. popular urban and rural mobilization in the struggle Between 1995 and 2009, Brazil experienced the to reform the structure of the Brazilian State in favour longest period of political stability in its modern history, of the less privileged classes. Conservative groups and largely thanks to the macroeconomic equilibrium the middle-class resisted the changes and supported achieved by the government of President Fernando the military uprising of 1964. The military government Henrique Cardoso and consolidated during the mandate phase began in 1964 with a provisional project, but it of President Luís Inácio da Silva. subsequently defined long-term policies as from 1969. Table 3 provides a summary of the results obtained During many of the military governments of that period, with the ins, indicating the periods of greatest instability there was organized resistance in the form of guerilla (1889-1894, 1895-1930, 1931-1945, 1961-1970 and 1985- movements which, despite not putting military rule at 1994) and the respective means and standard deviations of risk, represented a factor of political unrest. During the the ins. This shows that the greatest instability occurred mandate of Ernesto Geisel, the military began a lengthy between 1961 and 1970, followed by the period 1889- and costly process of transition to democracy. 1894, and the process of democratic transition. Thus, Democratic transition (1985-1994). This period apart from the period of the First Republic, political was complex. The death of President-elect Tancredo instability is a characteristic of modern Brazil. Neves and the swearing-in of Vice President José Sarney meant that, from the outset, the New Republic displayed great fragility in facing the economic problems 6 The 11 finance ministers were: Francisco Oswaldo Neves Dornelles, inherited from the last military government. In addition Dílson Domingos Funaro, Luiz Carlos Bresser Gonçalves Pereira and to economic difficulties, the new government had to Maílson Ferreira da Nóbrega (in the government of José Sarney); implement reforms to democratize the State. In 1988, Zélia Maria Cardoso de Mello and Marcílio Marques Moreira (in the government of Fernando Collor), and Paulo Roberto Haddad, Eliseu Brazil’s seventh constitution was approved, the most Resende, Fernando Henrique Cardoso, Rubens Ricupero and Ciro democratic of its history, which incorporated the “Getulista Ferreira Gomes (in the government of Itamar Franco).

TABLE 3 Brazil: values of the political instability index (ins) during the main episodes of instability, 1889-1994 Period Mean/standard deviation of the ins Episodes of instability 1889-1894 0.600/0.278 • Deodoro da Fonseca dissolved Parliament • Floriano Peixoto took office and did not call new elections • Constitutional military uprisings against Floriano Peixoto 1895-1930 0.132/0.217 • Federalist War in Rio Grande do Sul • War of Canudos • Copacabana Fort Revolution • Coluna Prestes revolt 1931-1945 0.194/0.413 • Civil and military coup led by Getúlio Vargas • Constitutions of 1934 and 1937 • Dissolution of Congress and declaration of the Estado Novo • Paulista War of 1932 • Military coup against Getúlio Vargas in 1945 1961-1970 0.906/0.430 • Resignation of Jânio Quadros • João Goulart took office • Frequent changes of more than 50% of the Cabinet • Public demonstrations for and against the reforms of the State • Coup d’état of 1964 • Amendment of the Constitution (five times) • Promulgation of Institutional Act No. 5 (AI-5) • Presence of armed insurgent movements

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 181

Table 3 (conclusion) Period Mean/standard deviation of the ins Episodes of instability 1985-1994 0.359/0.536 • Death of President-elect Tancredo Neves • Impeachment of President Fernando Collor • Continuous changes of Finance Minister • Six failed economic stabilization plans

Source: prepared by the authors.

3. Statistical correlation between the ins and a The analysis of table 4 was performed in decreasing selection of economic variables order of the indicators chosen. A negative relation was found between political instability and output. In the This subsection describes the indicators of statistical case of longer periods, that correlation does not seem correlation between the ins and certain economic very significant. Nonetheless, considering the period variables, with the aim of assessing whether each indicator since 1950, the correlation is significant, and the three is related to other variables in the logically expected information sources display similar results. The difference direction. In addition, the intensities of the calculated in correlation between the two periods shows that, in the correlations are evaluated. years of Brazil’s modernization, political instability had The economic indicators chosen were the level of more pronounced effects in terms of reducing output. A gdp, investment, the terms of trade, inflation, international more detailed and in-depth analysis of that phenomenon crises, the fiscal deficit, and population. As the indicator would go beyond the scope of this article; but it should produced spans a lengthy time series, covering the period be noted that the phenomenon is reflected in the three 1889-2009, various sources of information were used, information sources. with series covering a large number of years. A negative correlation can also be discerned between For the output series, the study by Maddison (2008) investment and political instability, which is almost was used, which contains Brazilian gdp data from 1870 nonexistent in the ibge-Histórico series considering to 2007. gdp data were also obtained from the Brazilian almost the entire period; but that situation changes when Geographical and Statistical Institute (ibge)-Histórico, only the more recent period between 1950 and 2009 is which contains data between 1900 and 2007. Apart considered. The behaviour of investment in Brazil in from that information, ibge-Histórico provides data the two subperiods is interesting. From 1900 to 1950, on investments and the terms of trade in those periods. the average investment rate was 10.5% of gdp, with a Another information source was the study by Heston, maximum of 20.3% of gdp and a minimum of 4.21%. Summers and Aten (2008), which includes data on output Between 1950 and 2007, the average investment rate and investments from 1950 to 2007, identified in table was 18.4%, with extreme values of 12.8% and 26.9% of 4 as Penn World Tables (pwt). Data on inflation and gdp. In the most recent period, investment represented international crises was taken from Rogoff and Reinhart a major share of aggregate demand and, therefore, its (2011) on Brazil as from 1830. In the case of the fiscal behaviour should more closely match variations in output deficit, the 1900-2007 series available on the Ipeadata and the domestic and external shocks to the Brazilian website was used. The original sources of those data economy. This possibly explains the higher correlation are the Ministry of Finance. The fiscal deficit indicator that exists between investment and political instability was used to calculate the deficit as a percentage ofgdp in more recent periods. estimated by ibge-Histórico. It is interesting to evaluate the correlation between Having different sources for a given variable, as in the the international terms of trade and political instability, case of output, made it possible to compare the different because negative shocks on the international markets for estimations and thus obtain more robust conclusions. Brazil’s export products serve as one of the channels Table 4, below, shows the correlation coefficients through which political instability is transmitted. That and highlights the different information sources. As the assessment was made using an indicator of changes Heston, Summers and Aten (2008) data start in 1950, in the terms of trade in a given year in relation to the it was decided to present the correlation coefficients previous one. Account was also taken of the volatility from the other sources, particularly those of Maddison of the terms of trade, which, for a given year, is equal (2008), ibge-Histórico, and Rogoff and Reinhart (2011) to the standard deviation of the terms of trade index in in comparable periods. the five previous years, divided by the average terms of

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 182 CEPAL REVIEW 114 • DECEMBER 2014

TABLE 4 Brazil: correlation coefficient between theins and a selection of economic variables, 1889-2007 Sources/variables Years Number of years Correlation coefficient gdp Maddison 1889-2007 118 -0.1771** ibge-Histórico 1901-2007 106 -0.1431*** Penn World Tables (pwt) 1950-2007 57 -0.4475* Maddison 1950-2007 57 -0.4698* ibge 1950-2007 57 -0.4611* Investment ibge-Histórico 1901-2007 106 -0.0154 ibge-Histórico 1950-2007 57 -0.2447 pwt 1950-2007 57 -0.1315 Terms of trade Terms of trade variation 1905-2007 102 -0.0403 Terms of trade volatility 1905-2007 102 -0.1848*** Terms of trade volatility 1950-2007 57 -0.1619 World output 1889-2007 118 -0.147*** World output 1950-2007 57 -0.4124* Inflation Variation in prices 1889-2007 118 -0.045 Variation in prices 1950-2007 57 -0.073 Financial crises Financial crises 1889-2007 118 0.1371*** Financial crises 1950-2007 57 0.2109*** Federal fiscal deficit Federal fiscal deficit 1900-2007 107 -0.0807** Federal fiscal deficit 1950-2007 57 -0.0691 Government income 1900-2007 107 -0.2271* Government income 1950-2007 57 -0.3717* Government expenditure 1900-2007 107 -0.2082* Government expenditure 1950-2000 57 -0.3596* Population Population 1889-2007 118 -0.148*** 1950-2007 57 -0.412* Source: prepared by the authors, on the basis of A. Maddison (2008), The World Economy: Historical Statistics, Paris, oecd Development Centre, 2008 [online] http://www.ggdc.net/maddison/; A. Heston, R. Summers and B. Aten, “Penn World Tables Version 6.1”, Center for International Comparisons at the University of Pennsylvania, 2008 [online] http://pwt.econ.upenn.edu/, and K. Rogoff and C. Reinhart, “From financial crash”, American Economic Review, vol. 101, No. 5, Nashville, Tennessee, American Economic Association, 2011. ins: political instability index. *Statistical significance at 1%; **statistical significance at 5%; ***statistical significance at 10%. trade of the five previous years. Thus, terms-of- trade of-trade volatility has a more significant effect; and, volatility in a given year reflects the variability of the doubtless, the economic activity level of the most terms of trade in the five preceding years, for which important countries is also clearly correlated with reason the volatility series begins in 1905. The definition political instability. of volatility and the data were obtained from the ibge- The correlation between the ins and inflation Histórico database. Another measure considered as an has the expected sign, but it is small in value and not alternative indicator of Brazil’s international economic statistically significant. As the Brazilian economy has situation is the level of economic activity in the most a very long track record of living with inflation, it important countries, namely the average gdp per capita would be somewhat risky to assert that the Brazilian of the United States, Germany, the United Kingdom and economy displayed political instability in the years of Japan. This indicator was obtained from the Maddison highest inflation. It would be more appropriate to say (2008) database. All of the indicators suggest that the that a political crisis was caused by inflation in years variation in the international terms of trade is only containing hyperinflationary episodes, in other words, weakly correlated with the ins. In contrast, terms- in the period 1988-1994 in the Brazilian case.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 183

The indicator of financial crises produced by Rogoff If there was a deficit, the fiscal deficit indicator was and Reinhart (2011) refers to four types of financial crisis: negative. Bearing in mind that the greater the political a crisis of the exchange rate and high inflation, cessation instability the greater is the estimated index, there of payment on the government’s external and domestic should be a negative relation between the two indicators. debts, stock market crashes and banking crises. Each of Table 4 reports a negative correlation between the fiscal these crises is evaluated as a categorical variable, taking deficit and political instability, which weakened slightly the value 1 when the phenomenon is present in a given between 1950 and 2007. The variation between the year and zero otherwise. Lastly, the authors establish a two periods reflects the fact that Brazil systematically synthesis variable by adding together the different types recorded fiscal deficits from 1900 to 1950. Fiscal of crises defined. Table 4 uses that synthesis variable balance was only attained in nine of those years, and which summarizes the four types of financial crisis. In there were deficits in the other 41 years. Between 1950 principle, there should be a positive correlation between and 2009, however, the opposite was the case: only the financial crisis indicator and the ins index, the results 20 deficits in that period, and just three since 1976. This of which are shown in table 4. As is the case with other was the result of the counterinflationary policies applied indicators used in table 4, the effects of financial crises in those years, based on fiscal rigour. The correlations are more closely correlated with political instability between public expenditure and government income, since 1950. The average of the financial crises between on the one hand, and political instability, on the other, 1889 and 2007 is 1.44. The most significant feature of are clearer. the Rogoff and Reinhart (2011) series is that the series Lastly, the correlation between population and average rises from 0.95 in the period 1889-1950 to 1.96 the ins is calculated, to analyse whether population in the subsequent period. This means that financial crises growth in the country, with all of the phenomena that in Brazil have been far deeper in more recent years. The that implies, is related to political instability. The result maximum value of the financial crisis indicator was 6 in is that there is in fact a negative correlation between 1986, according to the financial crisis index estimated population and the ins. by Rogoff and Reinhart (2011). In that year, all types The calculation of the correlation indicators of crisis defined by the authors were present in Brazil. presented above leads to the following conclusions: (i) It is highly significant that, in the period running from the estimated index and the chosen economic variables the external debt crisis in Mexico in 1982 to the launch are correlated, and in the expected direction; (ii) those of the Real Plan in Brazil in 1994, the financial crisis results are robust, because the indicators calculated using indicator was 3.9, the highest level in Brazil’s financial data from different sources produce similar results, and history for more than a century. (iii) the correlation is stronger since 1950, which was Another indicator used was the correlation between the period in which the country began a modernization the ins and the fiscal deficit as a percentage of gdp. process and acquired greater productive complexity.

VI Final thoughts

In the various lines of research, it is important to take construct an ins for Brazil for the period 1889-2009, account of the interaction that exists between social, based on the principal component statistical technique. economic and political variables. When seeking to The index produced is deemed to capture the understand a country’s political and economic trajectory, main cycles of political instability in Brazil’s history. this kaleidoscopic view may be particularly necessary Moreover, the analysis of the series shows that those to explain the causes of the successes (and the failures) cycles correspond only to certain years, so, Brazil is a of its development process. country with a low level of political instability. As many studies on the Brazilian economy ignore The authors consider that this research makes a this complex historical interaction, mainly because positive contribution by constructing an ins for Brazil there is no single variable that encompasses all of these for the first time, which could be used in various studies dimensions consistently, the present study set out to to fill a historical and empirical gap in the literature.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 184 CEPAL REVIEW 114 • DECEMBER 2014

ANNEX Sources of information on the variables

A.1 Variable: Interruption of constitutional No account was taken of changes in the ministerial mandate (nc) cabinet of the following provisional governments: • Governmental Junta (1930), from 24 October 1930 To quantify this variable, the following criteria and to 3 November 1930. information sources were used: • José Linhares (1945), from 29 October 1945 to 31 (i) Characteristics of the period of the constitutional January 1946. mandate for the President and rules on his • Carlos Luz (1955), from 8 November 1995 to 11 replacement, set out in detail in annex A.3. November 1955. (ii) Different events that represent the interruption of • Nereu Ramos (1955), from 11 November 1955 to the President’s constitutional mandate (nc). 31 January 1956. • 1889: coup d’état by Deodoro da Fonseca against • Provisional governments of Ranieri Mazzilli of the Emperor Dom Pedro II. 1961 (25 August 1961) and 1964 (from 2 April • 1892: coup d’état by Floriano Peixoto against 1964 to 15 May 1964). Deodoro da Fonseca. • Provisional governmental junta (1969), from 31 • 1918: death of President-elect Rodríguez Alves, August 1961 to 30 October 1969. who never took office. For each government, the name of the ministers • 1930: civil and military uprising of Getúlio Vargas. was determined and the date on which they began and • 1937: Getúlio Vargas declares himself dictator. ended their functions, as well as a number of ministers • 1946: Getúlio Vargas is deposed. there have been since the foundation of the Republic. For • 1954: suicide of Getúlio Vargas. example, Rui Barbosa was the first Minister of Finance • 1961: resignation of Jânio Quadros. of Brazil, in the government of Deodoro da Fonseca, • 1964: military coup. and was thus assigned No. 1, whereas Guido Mantega, • 1969: death of Costa Silva and his unconstitutional Finance Minister in the government of President Luís replacement by a military junta. Inácio Lula da Silva is No. 71. This listing made it possible • 1985: death of President-elect Tancredo Neves. to precisely identify all of the ministers and ministries • 1992: constitutional overthrow of President existing during each presidential term. The list of ministries Fernando Collor de Mello. excluded the Ministry of De-bureaucratization, which was short lived, created on 18 July 1979 and abolished A.2 Change of 50% of the composition of the on 14 February 1986. Nor did it include the Ministry of cabinet (mg) Administrative Reform created during the mandate of President Fernando Collor de Mello, which was discarded To generate quality statistics with this variable, it is by his successor, President Itamar Franco. necessary to know the number of ministers in all periods If the number of changes of minister in a given studied and their names; in addition to the number of year exceeded half of the total number of ministries, a times those changes exceeded half of the number of situation of political instability was inferred to exist in ministers in a given year. that year. For the purposes of this study, replacement by To establish the number of ministries, a list was an interim minister was not considered as a change of prepared in creation date order, indicating changes in ministers. If, in any one ministry, there were two or more name and the year of extinction. This information is changes in a given year, those changes are counted as shown in table A.2. separately. In the cases of President Fernando Henrique These data were used to prepare a list of the number Cardoso and President Luís Inácio Lula da Silva, both of of ministries that operated under each President of Brazil whom served a second term, the changes in each mandate from 1889 to 2009, in other words the 34 people who were considered separately. In other words, in the year officially took office as President of the Republic. The in which the President took office for a new mandate, list does not include the governmental juntas of 1930 the change of minister between the end of the first and 1969, or the three presidents-elect who never took mandate and the start of the second was not considered office (Júlio Prestes, Rodríguez Alves and Tancredo as a change of minister for the purposes of this study. Neves), the first of these because of the military uprising Lastly, changes in secretariats and other institutions in 1930, and the other two because they died. of ministerial status were not counted. The only exception

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 185

TABLE A.2 Brazil: ministries with creation dates between 1808 and 2009

Ministry Year of creation 1 Ministry of Finance (mf) 1808 2 Ministry of Justice (mj) 1822 3 Ministry of Foreign Relations (mre) 1852 4 Ministry of Agriculture and Food Supply (ma) 1909 5 Ministry of Transport a (mt) 1861 6 Ministry of Marine Affairs b (mm) 1891 7 Ministry of the Army b (me) 1891 8 Ministry of Work and Employment (mte) 1930 9 Ministry of Aeronautics b (maer) 1941 10 Ministry of Health (msau) 1953 11 Ministry of Development, Industry and Foreign Trade c (md) 1960 12 Ministry of Mines and Energy (mme) 1960 13 Ministry of Planning (mpl) 1962 14 Ministry of Communications (mc) 1967 15 Ministry of the Environment (mma) 1973 16 Ministry of Social Welfare (mps) 1974 17 Ministry of Culture (mcul) 1985 18 Ministry of Science, Technology and Innovation (mtct) 1985 19 Ministry of Tourism (mtur) 1992 20 Ministry of Sports (mes) 1995 21 Ministry of National Integration (mint) 1999 22 Ministry of Agricultural Development (mda) 1999 23 Ministry of Defence (md) 1999 24 Ministry of Cities (mcidades) 2003 25 Ministry of Aquaculture and Fisheries (mpesc) 2003 26 Ministry of Social Development Fight against Hunger (mfome) 2004

Source: prepared by the authors. a This ministry has had several names. 1860-1891: State Secretariat of Agribusiness, Commerce and Public Works; 1891-1906: Ministry of Industry, Roads and Public Works; 1906-1967: Ministry of Roads and Public Works; 1967-1990: Ministry of Transport; 1990-1992: Ministry Infrastructure; 1992-1992: Ministry of Transport and Communications; and 1992-2012: Ministry of Transport. b Abolished under a Complementary Law 97 of 10 June 1999, which created the Ministry of Defence. c Ministry abolished between 1990 and 1992. to that rule was the Ministry of the Casa Civil (Office The key points in each Constitution or constitutional of the Chief of Staff of the Presidency of the Republic), amendment were as follows: which was created on 3 November 1930 as the Secretariat • Presidential mandate according to the Constitution, of the Presidency of the Republic and became the Casa in number of years. Civil on 1 December 1938. This decision was taken • Processes for replacing the President if the President- because the Casa Civil Minister is an institution of great elect cannot fulfil the constitutional mandate owing political importance in Brazil, with functions similar to death or constitutional dismissal. to those of prime minister in a parliamentary regime.7 • The rules through which regimes of exception, arising as a result of the use of force, regulated A.3 Changes in the Constitution (mc) its presidential succession system. In particular, this was the case of the military governments of The analysis of this point is relevant for deciding 1964 to 1985. whether the presidential mandate has been interrupted in In relation to the issues mentioned, the situation accordance with the Constitution. Brazil has had seven depended on each Constitution: constitutions: 1822, 1891, 1934, 1937, 1946, 1967 and • Constitution of 1891: the presidential term lasted 1988; and there have been eight constitutional reforms: four years, and the elected president could not 1927, 1961, 1963, 1964-1967, 1967, 1968, 1969 and 1979. stand for a new mandate. In the event of death or resignation of the President, the Vice President took office only until a new vote was held, instead of 7 The result of this uprising is shown in a 70-page annex which, owing to space restrictions, is not included in this article, but is competing the mandate, as happens now (Brazilian available on request. Constitutions of 1891 and 1957).

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 186 CEPAL REVIEW 114 • DECEMBER 2014

• Constitution of 1934: the mandate of the President • Paulista Revolution of 1922. was four years, and the replacement lasted long • Constitutionalist Revolution of 1932. enough to organize new elections (Arruda and • Communist Uprising (Intentona) of 1935. Caldeira, 1986). • Caparão guerrilla war of 1967. • Constitution of 1937: the mandate was six years • Araguaia guerilla war between 1967 and 1974. with the possibility of re-election (Arruda and Caldeira, 1986). Source: J. Schulz, O Exército na política: origens da intervenção • Constitution of 1946: the presidential mandate was militar, 1850-1894, São Paulo, Editora da Universidade de São Paulo (edusp), 1994; and E. Seidl, “A formação de um exército à five years (Montellato, Cabrini and Catelli Jr., 2000). brasileira: lutas corporativas e adaptação institucional”, História, • Constitution of 1966: the presidential mandate was vol. 29, No. 2, São Paulo, 2010. five years (Montellato, Cabrini and Catelli Jr., 2000). • Constitutional amendment of 1961: thange from a The following events referred to earlier were excluded: presidential to a parliamentary system. (i) The Vaccine Revolt: excluded because it did not • Constitutional amendment of 1963: thange from a have a defined political objective. It represented parliamentary to a presidential system. popular resistance to the methods of application and • Constitutional amendment of 1969: treation of a generalization of immunization in the city of Rio military junta which replaced President Costa Silva. de Janeiro. This initiative involved the destruction • Institutional Act No. 1: tonducted a purge of of homes and shanty towns, which caused a climate opposition politicians and citizens; governed of popular resistance. elections between 1964 and 1967. (ii) The Revolt of the Lash: excluded because it involved • Institutional Act No. 2: dissolved the existing parties acts of indiscipline in the Brazilian navy, in response to and in practice established the two-party system the brutalities imposed by that institution’s hierarchy. between 1964 and 1967. • Institutional Act No. 3: instituted the holding of direct A.5 Strikes (greve) elections for the state governments. The mayors (prefeitos) of capital cities and “municipalities Statistics on strikes are based on Noronha (2009) for the of the national security area” where henceforth years 1978-2007, and on Simão (1981) for earlier years. appointed by the governors in 1964. • Institutional Act No. 4: compelled Congress to vote on the draft Constitution of 1967. TABLE A.5 Brazil: number of strikes, 1888-2002 • Institutional Act No. 5: dissolved Congress, suspended constitutional guarantees and authorized Period No. of strikes per year the executive to legislate on all matters in 1969. 1888-1890 2 • Constitutional amendment: repeal of Institutional 1901-1914 9 1914-1929 8 Act No. 5, 1979. 1930-1936 12 • Constitution of 1988: four-year presidential term: 1937-1944 1 in the event of the death of the President, the vice 1945-1964 43 1965-1968 13 president took office until the end of the mandate. 1969-1977 0 Source: http://www.duplipensar.net/dossies/historia-de las-eleicoes. 1978-1984 214 1985-1989 1 102 1990-1992 1 126 A.4 Civil wars and acts of organized political 1993-1994 842 1995-1998 865 violence recorded in the history of Brazil (gc) 1999-2002 440

Source: E.G. Noronha, “Ciclo de greves, transição política e The civil wars or acts of organized violence recorded estabilização: Brasil, 1978-2007”, Lua Nova, No. 76, São Paulo, in Brazilian history are as follows: 2009; A. Simão, Sindicato e Estado. Suas relações na formação do • War of Canudos between 1896 and 1897. proletariado de São Paulo, São Paulo, Editora Ática, 1981. • Federalist Revolution between 1893 and 1895. • Vaccine Revolt of 1904. To evaluate the effect of the strikes, this variable • Revolt of the Lash of 1910. was normalized on the interval [0.1], using the value • Contestado War between 1912 and 1916. of each year in relation to the maximum value in the • Copacabana Fort Revolution of 1922. period 1990-1992.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 187 ins 0.000 0.000 1.319 0.267 0.442 0.000 0.000 0.000 0.996 1.585 0.000 0.170 0.000 -0.005 -0.001 -0.005 -0.001 -0.005 -0.001 -0.005 -0.005 greve 0.038 0.001 0.011 0.038 0.011 0.001 0.001 0.038 0.011 0.001 0.001 0.038 0.011 0.038 0.001 0.011 0.038 0.001 0.011 0.038 0.001 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 gc 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 mc 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 nc 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 mg Year 1942 1936 1949 1931 1943 1937 1950 1932 1944 1938 1951 1933 1945 1939 1934 1946 1940 1935 1947 1941 1948 ins 0.170 0.170 0.896 0.170 0.591 0.442 0.592 0.442 0.170 -0.001 -0.001 -0.001 -0.001 -0.011 -0.001 -0.001 -0.001 -0.001 -0.001 -0.001 -0.001 greve 0.007 0.007 0.007 0.008 0.007 0.007 0.007 0.008 0.078 0.007 0.011 0.008 0.007 0.007 0.008 0.007 0.007 0.008 0.007 0.007 0.007 0 1 0 1 1 1 0 0 0 0 0 1 0 0 1 0 0 1 0 0 0 gc 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 mc 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 nc 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 mg Year 1915 1928 1910 1922 1916 1929 1911 1923 1917 1930 1912 1924 1918 1913 1925 1919 1914 1926 1920 1927 1921 f ins 0.443 0.593 0.171 0.272 0.271 0.272 0.998 0.272 0.864 0.170 0.000 0.443 0.000 0.000 -0.001 -0.001 -0.001 -0.001 -0.001 -0.001 -0.001 e 0.002 0.008 0.002 0.008 0.002 0.008 0.002 0.008 0.002 0.008 0.002 0.008 0.002 0.002 0.008 0.002 0.002 0.008 0.002 0.008 0.002 greve d 1 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 gc Matrix of data and results, 1889-2009 Matrix of data and results, c 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 mc b 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 nc results a 0 0 0 0 0 1 1 1 0 1 0 1 1 0 0 1 0 0 1 0 0 mg Data and

A.6. A.6 TABLE Year 1907 1889 1901 1895 1908 1890 1902 1896 1909 1891 1903 1897 1892 1904 1898 1893 1905 1899 1894 1906 1900

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT 188 CEPAL REVIEW 114 • DECEMBER 2014 ins 0.171 0.219 0.168 -0.053 -0.053 -0.053 -0.053 -0.104 -0.053 -0.053 -0.104 -0.053 -0.104 -0.053 -0.053 -0.053 greve 0.391 0.391 0.748 0.391 0.391 0.768 0.391 0.391 0.768 0.391 0.391 0.768 0.391 0.391 0.768 0.391 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 gc 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 mc 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 nc 0 0 1 0 0 0 0 0 0 0 1 0 0 0 1 0 mg Year 2005 1999 1994 2006 2000 1995 2007 2001 1996 2008 2002 1997 2009 2003 1998 2004 ins 0.171 0.729 0.460 0.171 0.171 0.139 0.000 0.139 0.000 0.865 0.000 -0.135 -0.026 -0.026 -0.026 -0.026 -0.026 -0.026 -0.132 -0.026 -0.135 greve 1.000 0.190 0.190 0.000 1.000 0.979 0.190 0.748 0.000 0.979 0.190 0.000 0.979 0.190 0.000 0.979 0.190 0.000 0.979 0.190 1.000 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 gc 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 mc 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 nc 0 0 1 0 0 1 0 1 0 0 1 0 0 1 0 0 0 0 0 0 0 mg Year 1978 1973 1992 1985 1979 1993 1974 1986 1980 1975 1987 1981 1976 1988 1982 1977 1989 1983 1990 1984 1991 f ins 0.171 1.314 0.171 0.267 0.996 0.267 0.588 0.724 0.267 0.267 0.896 1.314 0.896 0.267 1.490 0.993 0.171 -0.005 -0.005 -0.005 -0.005 e 0.038 0.038 0.000 0.038 0.038 0.000 0.038 0.012 0.038 0.038 0.012 0.038 0.038 0.012 0.038 0.038 0.012 0.038 0.000 0.038 0.000 greve d 0 0 1 0 0 1 0 0 0 0 0 0 0 1 0 0 1 0 1 0 1 gc c 0 0 0 1 0 0 0 1 0 0 1 0 0 1 1 0 1 0 1 1 0 mc b 0 0 0 1 0 0 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 nc a 0 0 0 0 0 1 1 1 0 0 1 1 0 0 0 0 1 0 0 1 0 mg prepared by the authors. mg indicates changes in 50% of the ministerial cabinet. constitutional mandate. nc refers to interruptions of the President’s mc reflects changes in the Constitution. acts of violence. and organized wars gc refers to civil refers to strikes. greve of political instability in Brazil. ins : index

Table A.6 (conclusion) Table Year a b c d e f 1952 1971 1964 1958 1972 1953 1965 Source: Source: 1959 1954 1966 1960 1955 1967 1961 1956 1968 1962 1957 1969 1963 1970

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT CEPAL REVIEW 114 • DECEMBER 2014 189

Bibliography Abreu, M. de Paiva (2000), “The Brazilian economy, 1928-1980”, cesifo Working Paper Series, No. 2249, Munich, cesifo Texto para Discussão, No. 433, Rio de Janeiro, Department of Group Munich. Economics, Catholic University of Rio de Janeiro. Fausto, B. (1996), História do Brasil, São Paulo, Editora da (1999), O Brasil e a economia mundial, 1930-1945, Rio Universidade de São Paulo (edusp). de Janeiro, Civilização Brasileira. Gurmu, S., P. Rilstone and S. Stern (1999), “Semiparametric estimation Aisen, A. and F. Veiga (2011), “How does political instability of count regression models”, Journal of Econometrics, vol. 88, affect economic growth?”, imf Working Papers, No. 11/12, No. 1, Amsterdam, Elsevier. Washington, D.C., International Monetary Fund. Heston, A., R. Summers and B. Aten (2008), “Penn World Tables Alesina, A. and G. Tabellini (1990), “A positive theory of fiscal Version 6.1”, Center for International Comparisons at the deficits and debt”, Review of Economic Studies, vol. 57, University of Pennsylvania [online] http://pwt.econ.upenn. edu/. No. 3, Oxford, Oxford University Press. ibge (Brazilian Geographical and Statitistical Institute) (n/d), "Séries Alesina, A. and others (1992), “Political instability and economic históricas e estatísticas" [online] http://seriesestatisticas.ibge. growth”, nber Working Papers, No. 4173, Cambridge, gov.br/default.aspx. Massachusetts, National Bureau of Economic Research (nber). Maddison, A. (2008), The World Economy: Historical Statistics, Annett, A. (2001), “Social fractionalization, political instability, and Paris, oecd Development Centre [online] http://www.ggdc. the size of government”, imf Staff Papers, vol. 48, No. 3, net/maddison/. Washington, D.C., International Monetary Fund. Montellato, A., C. Cabrini and R. Catelli Jr. (2000), História temática: Arruda, M. and C. Caldeira (1986), Como surgiram as constituições o mundo dos cidadãos, São Paulo, Ed. Scipione. brasileiras, Rio de Janeiro, Federation of Organizations for Muñoz, R. (2009), “Political instability and economic growth: the Social and Educational Assistance (fase). case of Venezuela (1983-2000)”, Documentos de Trabajo, Barro, R. (1996), “Determinants of economic growth: a cross-country Caracas, Andrés Bello Catholic University. empirical study”, nber Working Paper, No. 5698, Cambridge, Noronha, E.G. (2009), “Ciclo de greves, transição política e Massachusetts, National Bureau of Economic Research. estabilização: Brasil, 1978-2007”, Lua Nova, No. 76, São Paulo. (1991), “Economic growth in a cross section of countries”, Rogoff, K. and C. Reinhart (2011), “From financial crash”, American Quarterly Journal of Economics, vol. 106, No. 2, Oxford, Economic Review, vol. 101, No. 5, Nashville, Tennessee, Oxford University Press. American Economic Association. Bohn, F. (2003), “A note on corruption and public investment: the Rogoff, K. and A. Sibert (1988), “Elections and macroeconomic political instability threshold”, Economics Discussion Papers, policy cycles”, Review of Economic Studies, vol. 55, No. 1. No. 559, University of Essex. Schulz, J. (1994), O Exército na política: origens da intervenção Castelar, A. and others (2001), “Brazilian Economic Growth, 1900- militar, 1850-1894, São Paulo, Editora da Universidade de 2000: Lessons and Policy Implications”, Rio de Janeiro, draft. São Paulo (edusp). Constituição Brasileira de 1896 (1957), Planalto Central do Brasil, Seidl, E. (2010), “A formação de um exército à brasileira: lutas Coleção Documentos Brasileiros, Livraria José Olympio Editora. corporativas e adaptação institucional”, História, vol. 29, Cukierman, A., S. Edwards and G. Tabellini (1992), “Seigniorage No. 2, São Paulo. and political instability”, American Economic Review, vol. 82, Simão, A. (1981), Sindicato e Estado. Suas relações na formação do No. 3, Nashville, Tennessee, American Economic Association. proletariado de São Paulo, São Paulo, Editora Ática. Devereux, M.B. and J.F. Wen (1998), “Political instability, capital Solimano, A. (2003), “Governance crisis and the Andean Region: a taxation and growth”, European Economic Review, vol. 42, political economy analysis”, Macroeconomía del desarrollo No. 9, Amsterdam, Elsevier. series, No. 23 (LC/L.1860-P), Santiago, Chile, Economic Edwards, S. and G. Tabellini (1991), “Explaining fiscal policies and Commission for Latin America and the Caribbean (eclac). inflation in developing countries”, Journal of International United Nations publication, Sales No. E.03.II.G.33. Money and Finance, vol. 10, Amsterdam, Elsevier. Woo, J. (2005), “Social polarization, fiscal instability and Evia, J.L., R. Laserna and S. Skaperdas (2008), “Socio- growth”, European Economic Review, vol. 49, No. 6, political conflict and economic performance in Bolivia”, Amsterdam, Elsevier.

INDEX OF POLITICAL INSTABILITY IN BRAZIL, 1889-2009 • JAIME JORDAN COSTANTINI AND MAURICIO VAZ LOBO BITTENCOURT

CEPAL REVIEW 114 • DECEMBER 2014 191

cepal Review referees 2013 and January-August 2014

During this period, the following academics and researchers acted as referees for the cepal Review:

Jorge Abrahão Adolfo Figueroa Aldo Mascareño Manuel Agosin Jorge Friedman Carlos Massad Germán Alarco Guillermo Alex Fuentes Enlinson Mattos José Eduardo Alatorre Alvaro Fuentes Roxana Maurizio Erik Alencar de Figueiredo Sebastián Gallegos Andrés Mideros Roberto Alvarez Alvaro Gallegos Alejandra Mizala Angelo Alves Carrara André Golgher Sandro Eduardo Monsueto Adriana Amado Ivonne González Maristela Monteiro Verónica Amarante Pablo González Juan Antonio Morales Reynaldo Bajraj Alfredo Graham Nanno Mulder Nora Balzarotti Camila Gramkow Lucas Navarro Alberto Barreix Juan Carlos Guajardo Marcela Nogueira Ferrario Renato Baumann Graciela Gutman Cecilia Oreiro Janine Berg Niels Hermes Ramón Padilla Ricardo Bielschowsky Sebastián Herreros Ernesto Pastén Fabio Bittes Terra Daniel Heymann Michael Pedersen Bineswaree Bolaky Martín Hopenhayn Mauricio Pereira José Borello María Eugenia Ibarrarán Esteban Pérez Flor Brown Gonzalo Iberti Miguel Pérez Marisa Bucheli Ricardo Infante Ignacio Perrotini Alfredo Calcagno Dirk Jaspers Dirk Pilat Fernando Cantú Luis Felipe Jiménez Pilar Poncela Rodrigo Caputo Juan Pablo Jiménez Gabriel Porcile Pablo Carvallo Juan Jimeno Fernando Porta Simone Cecchini Valeria Jordán Pedro Quaresma de Araujo Alessandro Cigno Roberto Jovel Dagmar Raczynski Manlio Coviello Bernardo Kosacoff Márcia Rapini Gustavo Crespi Guillermo Larraín Guilherme Resende Renato da Fonseca Rodrigo Larraín Luis Riffo Eva Yamila da Silva Catela Alejandro Lavopa Juan Carlos Rivas Jaime de Pablo Guillermo Le Fort Virginia Robano Gaaitzen de Vries Daniel Lederman Mônica Rodrigues Robert Devlin Nuno Leitão Corina Rodríguez Armando Di Filippo Andrés López Adrián Rodríguez Eliseo Díaz Carlos Ludena Fernando Rojas José Díaz Gustavo Lugones Indira Romero Marco Dini Oscar Mac-Clure Gert Rosenthal Mario Duarte Canever Carlos Mallorquín Cecilia Rossel Hubert Escaith Xavier Mancero Renzo Rossini Ernesto Espíndola Sandra Manuelito Sebastián Rovira Andrés Fernández Cristián Mardones José Luis Ruiz Fernando Ferrari Ricardo Martínez Jaime Ruiz-Tagle Fernando Ferrari Filho Rodrigo Martínez Marco V. Sánchez Helder Ferreira de Mendonça Jorge Mario Martínez Gonzalo Saraví 192 CEPAL REVIEW 114 • DECEMBER 2014

Diego Saravia Rodrigo Travitzki Matías Vernengo Claudia Schatan Varinia Tromben Andrea Vigorito Daniel Sotelsek Rafael Urriola Francisco Villarreal Raimundo Soto Carlos Urzúa Pablo Villatoro Jan-Egbert Sturm Juan Pablo Valenzuela Rob Vos Guillermo Sunkel Raúl Vázquez López Jürgen Weller Xavier Tafunell Marco Vega Alicia Williner Joanilio R Teixeira Mario Velásquez Manuel Willington Mario Tello Cecilia Vera Víctor Tokman Sebastián Vergara

cepal Review is grateful for their invaluable assistance. CEPAL REVIEW 114

Guidelines for contributors to the CEPAL Review

In order to facilitate the submission, consideration and — Tables, figures and other elements should be inserted publication of articles, the editorial board of the cepal Review at the end of the text in the format in which they were has prepared the following information and suggestions to serve designed; they should not be inserted as “pictures”. as a guide for future contributors. Figures in Excel should include the corresponding worksheets. The submission of an article implies an undertaking by the author not to submit it simultaneously to other publications. — The location of tables and figures in the body of the The copyright to all articles published in the Review shall be article should be indicated in the appropriate place owned by the United Nations. as follows: Insert figure 1 Each article will be reviewed by the editorial board, which Insert table 1 may decide to submit it to external referees. — Tables and figures should include an explicit and complete Papers should be submitted in the original language (English, reference to their sources. French, Portuguese or Spanish). They will be translated into — Tables should indicate the period covered at the end of the the appropriate language by the relevant eclac services. title, and should indicate the units in which the data are Each article must be accompanied by a summary, no more than expressed in a subtitle (in italics and between brackets). 150 words in length, giving a brief description of its subject — The symbols referred to in the “Explanatory notes” matter and main conclusions. which appear on the page preceding the table of contents should be taken into account in the preparation of tables Each article must also carry three jel (Journal of Economic and figures. Literature) classification codes. The jel Classification Codes Guide is available online at www.aeaweb.org/jel/jel_class_ — Footnotes to tables and figures should be ordered system.php. consecutively using superscript lower-case letters. — Figures should be prepared bearing in mind that they Papers should be no longer than 10,000 words, including the will be printed in black and white. summary, notes and bibliography. Shorter papers will also be considered. Acronyms and abbreviations

Articles should be sent by e-mail to [email protected] in Word — Acronyms and abbreviations should not be used unless format. Papers in pdf format will not be accepted. absolutely necessary, in which case the full name should be written out the first time it occurs in the article. Style guide: Bibliography Titles should not be excessively long. — Bibliographical references should be directly related to the Footnotes content of the article and should not be excessively long. — It is recommended that footnotes be kept to a minimum. — At the end of the article, under the title “Bibliography”, all the necessary information should be included — It is recommended that footnotes not be used to cite accurately and in alphabetical order by author: name bibliographical references; such references should of author(s), year of publication, full name of article preferably be incorporated into the text. (if any) and publication (including any subtitle), city of — Footnotes should be numbered consecutively using publication, publisher and, in the case of a periodical, superscript Arabic numerals. month of publication.

Tables, figures and equations The editorial board of the Review reserves the right to make any necessary editorial changes in the articles, including their titles. — It is recommended that tables and figures be kept to a minimum, avoiding any redundancy with the text. Authors will receive a one-year courtesy subscription to the — Equations should be written using the Office equation editor Review, plus 30 offprints of their article in Spanish and 30 in “MathType”; they should not be inserted as “pictures”. English, at the time of publication in each language.

DECEMBER 2014 Publicaciones recientes de la CEPAL ECLAC recent publications www.cepal.org/publicaciones

Informes periódicos / Annual reports También disponibles para años anteriores / Issues for previous years also available • Estudio Económico de América Latina y el Caribe 2014, 222 p. Economic Survey of Latin America and the Caribbean 2014, 214 p. • La Inversión Extranjera Directa en América Latina y el Caribe 2013, 160 p. Foreign Direct Investment in Latin America and the Caribbean 2013, 152 p. • Anuario Estadístico de América Latina y el Caribe 2013 / Statistical Yearbook for Latin America and the Caribbean 2014, 238 p. • Balance Preliminar de las Economías de América Latina y el Caribe 2014, 92 p. Preliminary Overview of the Economies of Latin America and the Caribbean 2014, 92 p. • Panorama Social de América Latina 2014, 296 p. Social Panorama of Latin America 2013, 284 p. • Panorama de la Inserción Internacional de América Latina y el Caribe 2014, 148 p. Latin America and the Caribbean in the World Economy 2014, 140 p.

Libros y documentos institucionales / Institutional books and documents • El desafío de la sostenibilidad ambiental en América Latina y el Caribe: textos seleccionados de la CEPAL 2012-2014, 2015, 148 p. • Panorama Fiscal de América Latina y el Caribe 2015: Dilemas y espacios de políticas, 2015, 130 p. • La economía del cambio climático en América Latina y el Caribe: paradojas y desafíos del desarrollo sostenible, 2014, 98 p The economics of climate change in Latin America and the Caribbean: Paradoxes and challenges. Overview for 2014, 2014, 76 p. • Los pueblos indígenas en América Latina: avances en el último decenio y retos pendientes para la garantía de sus derechos, 2014, 410 p. • Pactos para la igualdad: hacia un futuro sostenible, 2014, 340 p. Covenants for Equality: Towards a sustainable future, 2014, 330 p. • Integración regional: hacia una estrategia de cadenas de valor inclusivas, 2014, 226 p. Regional Integration: Towards an inclusive value chain strategy, 2014, 218 p. Integração regional: por uma estratégia de cadeias de valor inclusivas, 2014, 226 p. • Reflexiones sobre el desarrollo en América Latina y el Caribe. Conferencias magistrales 2013-2014, 2014, 100 p. • Prospectiva y desarrollo: el clima de la igualdad en América Latina y el Caribe a 2020, 2013, 72 p. • Comercio internacional y desarrollo inclusivo: construyendo sinergias, 2013, 210 p. International trade and inclusive development: Building synergies, 2013, 200 p. • Cambio estructural para la igualdad: una visión integrada del desarrollo, 2012, 330 p. Structural Change for Equality: an integrated approach to development, 2012, 308 p. • La hora de la igualdad: brechas por cerrar, caminos por abrir, 2010, 290 p. Time for Equality: closing gaps, opening trails, 2010, 270 p. A Hora da Igualdade: Brechas por fechar, caminhos por abrir, 2010, 268 p. Libros de la CEPAL / ECLAC books 132 Neoestructuralismo y corrientes heterodoxas en América Latina y el Caribe a inicios del siglo XXI, Alicia Bárcena, Antonio Prado (eds.), 2014, 452 p. 131 El nuevo paradigma productivo y tecnólogico: la necesidad de políticas para la autonomía económica de las mujeres, Lucía Scuro, Néstor Bercovich (eds.), 2014, 188 p. 130 Políticas públicas para la igualdad de género: un aporte a la autonomía de las mujeres, María Cristina Benavente, Alejandra Valdés, 2014, 134 p. 129 Prospectiva y política pública para el cambio estructural en América Latina y el Caribe, Javier Medina Vásquez, Steven Becerra y Paola Castaño, 2014, 338 p. 128 Inestabilidad y desigualdad: la vulnerabilidad del crecimiento en América Latina y el Caribe, Juan Alberto Fuentes Knight (ed.), 2014, 304 p. 127 Global value chains and world trade: Prospects and challenges for Latin America, René A. Hernández, Jorge Mario Martínez-Piva and Nanno Mulder (eds.), 2014, 282 p. 126 Planificación, prospectiva y gestión pública: reflexiones para la agenda del desarrollo, Jorge Máttar, Daniel E. Perrotti (eds.), 2014, 250 p.

Copublicaciones / Co-publications • Decentralization and Reform in Latin America: Improving Intergovernmental Relations, Giorgio Brosio and Juan Pablo Jiménez (eds.), ECLAC / Edward Elgar Publishing, United Kingdom, 2012, 450 p. • Sentido de pertenencia en sociedades fragmentadas: América Latina desde una perspectiva global, Martín Hopenhayn y Ana Sojo (comps.), CEPAL / Siglo Veintiuno, Argentina, 2011, 350 p. • Las clases medias en América Latina: retrospectiva y nuevas tendencias, Rolando Franco, Martín Hopenhayn y Arturo León (eds.), CEPAL / Siglo XXI, México, 2010, 412 p.

Coediciones / Co-editions • Perspectivas económicas de América Latina 2015: educación, competencias e innovación para el desarrollo, CEPAL/OCDE, 2014, 200 p. Latin American Economic Outlook 2015: Education, skills and innovation for development, ECLAC,/CAF/OECD, 2014, 188 p. • Regional Perspectives on Sustainable Development: Advancing Integration of its Three Dimensions through Regional Action, ECLAC-ECE-ESCAP-ESCWA, 2014, 114 p. • Perspectivas de la agricultura y del desarrollo rural en las Américas: una mirada hacia América Latina y el Caribe 2014, CEPAL / FAO / IICA, 2013, 220 p.

Cuadernos de la CEPAL 101 Redistribuir el cuidado: el desafío de las políticas, Coral Calderón Magaña (coord.), 2013, 460 p. 101 Redistributing care: the policy challenge, Coral Calderón Magaña (coord.), 2013, 420 p. 100 Construyendo autonomía: compromiso e indicadores de género, Karina Batthyáni Dighiero, 2012, 338 p. 99 Si no se cuenta, no cuenta, Diane Alméras y Coral Calderón Magaña (coords.), 2012, 394 p.

Documentos de proyecto / Project documents • La economía del cambio climático en el Perú, 2014, 152 p. • La economía del cambio climático en la Argentina: primera aproximación, 2014, 240 p. • La economía del cambio climático en el Ecuador 2012, 2012, 206 p.

Cuadernos estadísticos de la CEPAL 42 Resultados del Programa de Comparación Internacional (PCI) de 2011 para América Latina y el Caribe. Solo disponible en CD, 2015. 41 Los cuadros de oferta y utilización, las matrices de insumo-producto y las matrices de empleo. Solo disponible en CD, 2013.

Series de la CEPAL / ECLAC Series Asuntos de Género / Comercio Internacional / Desarrollo Productivo / Desarrollo Territorial / Estudios Estadísticos / Estudios y Perspectivas (Bogotá, Brasilia, Buenos Aires, México, Montevideo) / Studies and Perspectives (The Caribbean, Washington) / Financiamiento del Desarrollo/ Gestión Pública / Informes y Estudios Especiales / Macroeconomía del Desarrollo / Manuales / Medio Ambiente y Desarrollo / Población y Desarrollo/ Política Fiscal / Políticas Sociales / Recursos Naturales e Infraestructura / Reformas Económicas / Seminarios y Conferencias. Revista CEPAL / CEPAL Review La Revista se inició en 1976, con el propósito de contribuir al examen de los problemas del desarrollo socioeconómico de la región. La Revista CEPAL se publica en español e inglés tres veces por año. CEPAL Review first appeared in 1976, its aim being to make a contribution to the study of the economic and social development problems of the region. CEPAL Review is published in Spanish and English versions three times a year.

Observatorio demográfico /Demographic Observatory Edición bilingüe (español e inglés) que proporciona información estadística actualizada, referente a estimaciones y proyecciones de población de los países de América Latina y el Caribe. Desde 2013 el Observatorio aparece una vez al año. Bilingual publication (Spanish and English) proving up-to-date estimates and projections of the populations of the Latin American and Caribbean countries. Since 2013, the Observatory appears once a year.

Notas de población Revista especializada que publica artículos e informes acerca de las investigaciones más recientes sobre la dinámica demográfica en la región. También incluye información sobre actividades científicas y profesionales en el campo de población. La revista se publica desde 1973 y aparece dos veces al año, en junio y diciembre. Specialized journal which publishes articles and reports on recent studies of demographic dynamics in the region. Also includes information on scientific and professional activities in the field of population. Published since 1973, the journal appears twice a year in June and December.

Las publicaciones de la CEPAL están disponibles en: ECLAC Publications are available in: www.cepal.org/publicaciones

También se pueden adquirir a través de: They can also be ordered through: www.un.org/publications

United Nations Publications PO Box 960 Herndon, VA 20172 USA

Tel. (1-888)254-4286 Fax (1-800)338-4550 Contacto / Contact: [email protected] Pedidos / Orders: [email protected] 114

REVIEW EDITORIAL BOARD

Osvaldo Sunkel Chairman

André Hofman José Antonio Alonso Director Oscar Altimir C Renato Baumann Miguel Torres Luis Beccaria Technical Editor Luis Bértola Luiz Carlos Bresser-Pereira Mario Cimoli www.eclac.org/revista John Coatsworth Robert Devlin Carlos De Miguel Ricardo Ffrench-Davis Daniel Heymann Martín Hopenhayn Akio Hosono Graciela Moguillansky Juan Carlos Moreno-Brid José Antonio Ocampo Carlota Pérez Gert Rosenthal Paul Schreyer Barbara Stallings Andras Uthoff Rob Vos

ECONOMIC COMMISSION FOR LATIN AMERICA AND THE CARIBBEAN

United Nations Publication • E1420359 • December 2014 • ISSN 0251-2920 • e-ISBN 978-92-1-057211-8 Copyright © United Nations 2014 • Printed in Santiago, Chile